# Tweets Curated tweets worth keeping. - **jxnlco** (2026-08-01): https://x.com/jxnlco/status/2083578220163809334 - @polynoamial @OpenAI i cant believe i used it to itemize a navan receipt... im sorry... - **jarredsumner** (2026-08-01): https://x.com/jarredsumner/status/2083542090953470212 - the spike in july: new fuzzer overrides syscalls and diffs bun’s behavior against node on identical inputs. a /loop in claude code drives it and sends bugs to robobun to fix https://t.co/DJl8b5HiDR - **polynoamial** (2026-08-01): https://x.com/polynoamial/status/2083467194663571701 - An internal version of Astra, @OpenAI’s next major model family, solved 10 major open problems in mathematics, quantum complexity, and theoretical computer science. We believe it will be a major step for scientific reasoning. https://t.co/iP6cyheZ7i https://t.co/jHuulDwV46 https://t.co/DfBOvi5O8F - **gdb** (2026-08-01): https://x.com/gdb/status/2083424030099656806 - openai team making git better for everyone https://t.co/Vb3BbmuM8b - **OaklandFireCA** (2026-08-01): https://x.com/OaklandFireCA/status/2083392608660005012 - Update #1) Incident Commander requested a 2nd Alarm. 50+ firefighters are currently on scene of veg fire near Sequoia Arena staging area. There have been 3 water drops from helicopters. Forward progress is stopped, most of fire is knocked down, crews checking for hot spots. https://t.co/padRK1WkEj https://t.co/skabw2VZ0S - **DhravyaShah** (2026-07-31): https://x.com/DhravyaShah/status/2083259863355015254 - Hats off to @mattzcarey for sending a PR for updating the mcp to the latest spec 🫡 https://t.co/94rsdv7O2b - **burcs** (2026-07-31): https://x.com/burcs/status/2083258460922110260 - i have no clue know why they didn't go with my announcement graphic... https://t.co/FfLTemDkRc https://t.co/eEDGPgrs4y - **michellechen** (2026-07-31): https://x.com/michellechen/status/2083256499430961306 - when we host a model on workers ai, we have to consider if we're going to be able to serve a model economically + what demand looks like + what capacity we have. testing models on @OpenRouter allows us to launch these things in market and gather data. if we have to tear a model down, your traffic gets routed to other providers. once we're confident in our deployment, then we bring it on platform for Workers AI users, where we commit to running the model for longer. it's been a joy to work with openrouter team to enable us to experiment with models quickly 🧡 - **dok2001** (2026-07-31): https://x.com/dok2001/status/2083253946412315083 - Fred just couldn’t wait until next week to launch. https://t.co/20iM7BqOXT - **FredKSchott** (2026-07-31): https://x.com/FredKSchott/status/2083252640994275779 - Introducing Flue 2 — Build dynamic agents that evolve over time. Flue is a TypeScript framework for building the next generation of agents, powered by the @pidotdev agent harness. The Flue 1.0 Beta was released last month & has now been downloaded 700,000+ times. And from all of that feedback, we realized something we'd missed. Agents have a familiar problem: composability. How do you compose together different agent capabilities in a way that feels expressive AND maintainable, especially as you scale up complexity? Last week, I teased: “What if React for Agents?” and yes, as you will see, the inspiration here is obvious. But UI devs and agent devs have some things in common, including the same problem that the React team solved almost a decade ago with... Hooks! There are now things that you can do in Flue 2.0 with hooks that are just fundamentally difficult in other frameworks: A Flue agent can enforce a multi-step workflow, earn new tools once certain conditions are met, or trade up to a bigger model when the work gets too hard. Handling metadata and context no longer requires custom framework APIs and configuration. In Flue, you can pass data to a tool the same way you pass arguments to a function. I'm really excited for you to try Flue 2.0 out yourself. It's a big, bold take on what agents could be capable of in the future. Read more: https://t.co/kW69RWJ00Q - **simeonGriggs** (2026-07-31): https://x.com/simeonGriggs/status/2083244161428279676 - @brandur You can create PlanetScale DB's through that same UI - **FredKSchott** (2026-07-31): https://x.com/FredKSchott/status/2083239731622760926 - @flueai can't stop until the boss is happy https://t.co/4bydmu2H6e - **flueai** (2026-07-31): https://x.com/flueai/status/2083238270687945103 - Introducing Flue 2 — Build dynamic agents that evolve over time, powered by our brand new Agent Hooks API. https://t.co/NjhWCTUl8q - **dok2001** (2026-07-31): https://x.com/dok2001/status/2083193896822055228 - . @KentonVarda was just ahead of his time again. There is still work to be done to make the technology easily by the rest of us. Capabilities security models are just not the mental model most people have for security. https://t.co/fqgmEovir7 - **thdxr** (2026-07-31): https://x.com/thdxr/status/2083182360816910786 - openai: well actually we did 100 hacks anthropic: we check again we did a million hacks openai: we do infinity hacks anthropic: we do infinity times infinity hacks https://t.co/XEf45FTBDO - **artman** (2026-07-31): https://x.com/artman/status/2083040643467587603 - Stacked PR’s are an outdated, inefficient solution to a problem that only exist because the root problem hasn’t been addressed in GitHub: review tools need to do a better job at help the reviewer work through large code changes. It’s a local maxima. https://t.co/0mfCFh5tT8 - **dok2001** (2026-07-31): https://x.com/dok2001/status/2083038219319566381 - Twice in nine days. OpenAI's models chained a zero-day to get out of an eval environment. Anthropic just found three incidents of the same shape. This is what capable models do. Every reachable path is an invitation. Credit to both for publishing. We rebuilt Cloudflare OS, our internal agent platform, assuming exactly this. Sandboxes have no network path out. Agents don't get every MCP tool up front, each workspace gets only what it needs. And instead of stopping for every approval until someone gives up and enables auto-approve, the platform simulates pending actions so the agent keeps working, then queues the real ones for review at the end. The app side goes further. Apps carry no auth code at all, the sandbox provides it, so there is no auth logic to get wrong. Multi-document apps are isolated per document, so a bug can't leak between separately shared items. Anyone viewing an app must have direct permission to each of its data sources, so a dashboard on sensitive data can't overshare. Actions need the user's own permission or an approval from someone who has it, so an app can't escalate its author's privileges. And everything an app does is logged, even when the service it calls doesn't support logging. Most companies are about to point agents at real systems. Build for what the models actually do. We will open source this next week! - **gakonst** (2026-07-31): https://x.com/gakonst/status/2083027609768878358 - @irvinebroque this + MPP is going to be overpowered im telling you - **gakonst** (2026-07-31): https://x.com/gakonst/status/2082991969224331747 - got nanocodex running on cf durable objects! cool! https://t.co/qhvVgvXY6c https://t.co/H9RdGPUyjs - **emollick** (2026-07-30): https://x.com/emollick/status/2082924870497681712 - New uses require new interfaces These are the three things I have been trying out to control Codex. Given how useful voice mode is, the Teenage Engineering Ting walky-talky has been a surprising favorite. The Codex micro is beautiful & fun, but too little info vs the Stream Deck https://t.co/m5YCUtBvC2 - **clairevo** (2026-07-30): https://x.com/clairevo/status/2082912432398475400 - I’m not a mom on vacation I’m a forward deployed parent - **gakonst** (2026-07-30): https://x.com/gakonst/status/2082882218612109461 - https://t.co/CIFYdlKCAH - **OpenAI** (2026-07-30): https://x.com/OpenAI/status/2082878156483219672 - We are committed to pushing the model frontier across cost efficiency, capability, and speed. Starting today, we are reducing prices for GPT-5.6 Luna by 80% and GPT-5.6 Terra by 20% , and offering a faster option for GPT-5.6 Sol in the API. Luna and Terra’s lower prices are reflected in how usage is counted in Codex and ChatGPT Work, so your usage goes further. - **vite_js** (2026-07-30): https://x.com/vite_js/status/2082871231523459218 - Vite 8.2 is out 🚀 • Top-level `input` option: No need to use `build.rolldownOptions` • Warnings for config features the native loader can't handle yet • PostCSS config type export for type-safe configs • Dev URLs now labeled by their interfaces https://t.co/xJR0baP0hd - **michellechen** (2026-07-30): https://x.com/michellechen/status/2082835159531897195 - teasing something new... https://t.co/GzsC2vUT55 https://t.co/vcQUUtrvX0 - **aninibread** (2026-07-30): https://x.com/aninibread/status/2082812453230919850 - if you're building with 🦜langchain or the 🪄ai sdk, there's now a native @CloudflareDev ai search integration for both. feed it your data, it indexes it, and you get grounded context for your agents w/o building the backend yourself. 👉 https://t.co/SjZUBu8gwa https://t.co/vJprY5GKlM - **mike_rosinsky** (2026-07-30): https://x.com/mike_rosinsky/status/2082643515624800613 - So much blood sweat and tears from the API team to get this up and running. But it’s here. e2e encrypted chat via the X API. And this is only the beginning of what we have planned for chat integrations. https://t.co/wfv59UEKip - **just_be_dev** (2026-07-30): https://x.com/just_be_dev/status/2082640539652895145 - @irvinebroque Oh, damn, looks like I’m writing another effect library tonight - **ibby_rips** (2026-07-30): https://x.com/ibby_rips/status/2082636847511990488 - @WilcoKr @irvinebroque hey! I work on the team. What exactly do you mean 👀 - **jarredsumner** (2026-07-30): https://x.com/jarredsumner/status/2082628139352780975 - @sorenbs @irvinebroque @bunjavascript Yeah. - **gakonst** (2026-07-30): https://x.com/gakonst/status/2082620966396641682 - i'll be there presenting the machine payments protocol with @brendan_j_ryan - DM Jeff as it'll be a rare moment where we preview the future https://t.co/m2VKnCxc8d - **darkmembo** (2026-07-29): https://x.com/darkmembo/status/2082606649408381228 - orange = better https://t.co/kViqIGofPZ - **charliermarsh** (2026-07-29): https://x.com/charliermarsh/status/2082603564921200775 - Codex CLI installs now go through our Cloudflare CDN by default instead of relying on GitHub Releases. Enjoy the improved reliability and install speed! (On my machine, it's 25% faster end-to-end.) - **leinwand** (2026-07-29): https://x.com/leinwand/status/2082586167598543145 - Networking engineers wanted: help build the most innovative network on the planet, spanning 330+ cities. If you want to work on something big, this is your chance to work at @Cloudflare and help build a better Internet: https://t.co/gRgY5ABiIT - **reed** (2026-07-29): https://x.com/reed/status/2082574994526425101 - FREEWAYS ARE BACK @WAYMO - **boristane** (2026-07-29): https://x.com/boristane/status/2082553165996781971 - founder used to mean something - **jjacky** (2026-07-29): https://x.com/jjacky/status/2082514913314947166 - many such cases https://t.co/pq97ssabuh - **mitch_troy** (2026-07-29): https://x.com/mitch_troy/status/2082513195357307158 - Out of the box, long-horizon agents struggle to accurately perform end to end work in the real economy (outside of coding) because those tasks are not easily verifiable, the data is hard to scale, and going from inputs to real outcomes can actually take many days. Even if you had a reliable way to verify outcomes at scale (and weren’t bothered by the multi-hour iteration loops), the sheer volume of decisions by the agent that occur in a multi-hour job makes it hard to know whether performing well will generalize to production. Over the last two years at @trybasis, we've been solving this problem by supervising the process our agents take to get to outcomes, rather than just looking at whether the outcome itself is correct. We think this is the key to building production agents at scale. It's what has allowed us to run agents in production that operate for hours, sometimes days, and reliably perform tasks like entire complex tax returns end to end. Today, alongside @braintrust, we're open sourcing a standard for defining, evaluating, and eventually rewarding agent behaviors. Thread below with all the details on how we’re scaling behaviors to close the loop for long-horizon agents. - **kevin_flansburg** (2026-07-29): https://x.com/kevin_flansburg/status/2082509689628340279 - kiMi iS sO exPenSivE! But you've been paying the same for worse performance from Sonnet for years. - **southpolesteve** (2026-07-29): https://x.com/southpolesteve/status/2082496282715643950 - Too many people trying to get AI to write whole codebase. Not enough using it for the boring shit nobody has time for: Improve your error messages. Build better on call dashboards. Fix the broken stack traces in Sentry. Clean up your old feature flags. Low stakes stuff AI is actually good at and you weren't going to get to anyways. - **MattieTK** (2026-07-29): https://x.com/MattieTK/status/2082484567529685131 - Integration testing on @Cloudflare Workers? We've just made it a lot easier for you. 🗑️ bin unstable_startWorker() 🎁 unwrap createTestHarness() See how much cleaner a simple test setup looks like. Read more: https://t.co/D02N2RdKbS https://t.co/wPnnp6a9e5 - **madbyk** (2026-07-29): https://x.com/madbyk/status/2082449206443188646 - A VERY Orange release - much love @Cloudflare https://t.co/ga49zbBqBP - **darkmembo** (2026-07-29): https://x.com/darkmembo/status/2082327952641974748 - just met with the CTO of a 9-figure company. off-hand quote: "I'm so tired of babysitting AWS. we benched cloudflare, it just scales -- we are building on you guys now." - **jeff_weinstein** (2026-07-29): https://x.com/jeff_weinstein/status/2082292674967880044 - @kevinyien and I have been chatting re: the product manager's role — one thought: (1) make sure the team ships (2) decide to ship something "surprisingly great" or "not obviously bad" if "surprisingly great" = yay else, ensure "not obviously bad" to earn trust, then ship again - **skeptrune** (2026-07-29): https://x.com/skeptrune/status/2082264691196461056 - extremely proud of this one s/o @rohandevs https://t.co/BwhxSquVqD - **thsottiaux** (2026-07-28): https://x.com/thsottiaux/status/2082241164850364555 - More opensource goodness. We have just released a CLI and TypeScript SDK for finding, validating, and fixing security vulnerabilities in your code. Scan repositories, review changes, track findings over time, and run security checks in CI. https://t.co/nkfTbw8p7b - **dsp_** (2026-07-28): https://x.com/dsp_/status/2082173429399142616 - Thrilled to announce the MCP 2026-07-28 release. It is one of MCP's biggest yet, built on 18 months of lessons learned. A few highlights: ★ MCP is now stateless, with semantics for multi-round-trip requests. Serving MCP just got much simpler and more scalable. ★ MCP now has extensions. MCP Apps, Tasks, Enterprise Managed Auth, and more. Domain-specific ways to use MCP, plus room for experimental additions. ★ Python, Typescript, C# (soon) SDKs released a v2.0.0 for the new spec with improved ergonomics! But the biggest highlight for me: this release is a true community effort. Individuals and companies alike — Anthropic, Google, Microsoft, OpenAI, and many others helped shape this specification. And there's much more. Full details: https://t.co/NHd9cQmPCu - **Vercantez** (2026-07-28): https://x.com/Vercantez/status/2082138839888589200 - https://t.co/pGb77Ubtdp - **grinich** (2026-07-28): https://x.com/grinich/status/2081923651436208218 - The 2 hardest problems in AuthZ right now are: 1) Permissions for multiplayer agents+humans 2) Permissions for non-deterministic agentic systems We have a solution at @WorkOS that I'm bullish about. Looking forward to sharing more soon. 👀 https://t.co/3UKTnPWtF6 - **andrewchen** (2026-07-28): https://x.com/andrewchen/status/2081916816000241712 - ai startups aren’t competing against other ai startups. They’re competing against: - the tracker sheet their team’s used for years - the 5 open tabs needed for a task - the copy-and-paste workflow nobody documented - the employee who “just does it” - ignorance, fear of change, laziness - **mateosfo** (2026-07-27): https://x.com/mateosfo/status/2081890669694751157 - Lots of folks wondering why this beautiful housing proposal in the Marina matters so much, I realize it's been a minute since we've discussed the outsized impact San Francisco NIMBYism has on US politics and economy. It's ... not small: A significant share of 12 - 36% of GDP. https://t.co/JBc5TTLDlh - **zeeg** (2026-07-27): https://x.com/zeeg/status/2081870681407639710 - one of the biggest gripes i have about tech (and startups) right now is... everyones trying to build the-everything-solution what i mean is: many startups should be building "features" that can slot in, maybe break out one day. say take all the components you need to make a coding agent work: the IDE, verification layers, sandboxes/vms, security, etc. but instead they're not doing that, and thus no one is achieving any breakout success. they're all just me-toos that dont work that well, and certainly not better than the other persons idea i dont really blame them as you are taught to swing big, and you're somewhat naively convinced you can right now. except whats happening is everyones blindly confident that they've found _the secret sauce_ (100% placebo btw) and they barely ever scratch the surface on average I'd agree with the wisdom that you should aim big, avoid being the feature, but this is an odd domain where the problems - each component - is so uncommonly difficult and valuable that we SHOULD be building those. and thats just in software dev! you see the same thing in how indie developers operate, and in open source, where instead of trying to lego together things to (truly) solve a really big problem, everyones pretending they can vibe code some slop with the way its going we're like six months away from people thinking SQL is too complicated and having their agents invent new graph stores... - **brian_scanlan** (2026-07-27): https://x.com/brian_scanlan/status/2081844295510667387 - Speed-learning Run Less Software: https://t.co/s9AdtHIAHX https://t.co/QY0WZSH9VN - **matthewcp** (2026-07-27): https://x.com/matthewcp/status/2081828600106336608 - I think 2 things are happening in parallel that cause this: 1. People look at the code less and less. This is especially true of newer developers. So there isn't the pull to try new things like there was previously. 2. Framework decision is made by fewer people than in the past. We're approaching a world where the people who build agents and write the system prompt are the ones who are most influencial over framework choice. - **DanielleMorrill** (2026-07-27): https://x.com/DanielleMorrill/status/2081815301176647872 - @tymrtn Cloudflare - **DanielleMorrill** (2026-07-27): https://x.com/DanielleMorrill/status/2081788653177864623 - the only way to win in SaaS now is to massively increase the scope do what you do, solve 10x the streamlining of processes for users, and make it all composable jumping to a whole new plane of abstraction is so hard with 100+ employees, new companies needed - **gokulr** (2026-07-27): https://x.com/gokulr/status/2081782731848224997 - From someone at @OpenAI who i respect greatly, about the force behind the recent excellence in Codex: "Tibo (@thsottiaux) is THE Guy. He’s the real deal." Keep crushing, Tibo! - **feross** (2026-07-27): https://x.com/feross/status/2081779456943816861 - I'm an engineer. I used to think sales was dirty. Cold calls, pushy tactics, all of it. The product should speak for itself. Running Socket changed my mind. I actually like sales now. Done well, sales is consultative. You help people understand a problem that matters, you're honest about where your product is strong and where it isn't, and you optimize for fit. That's not in tension with the engineering mindset. It's an extension of it. There are two ways to die as a company. Build a technically excellent product that nobody cares about. Or put sales in charge, underinvest in engineering, grow fast, and hit a ceiling when customers realize the product hasn't improved in five years. We've watched competitors do the second. The job is to keep engineering and sales both strong, and never let one dominate the other. I joined Anne Dwane on the Village Global podcast to talk about this, plus where Socket is today and why open source supply chain risk keeps getting worse. Full episode in the comments. - **harshil1712** (2026-07-27): https://x.com/harshil1712/status/2081775799925977113 - Oh well, I tried running it on @CloudflareDev! https://t.co/lLHF0Cx5Xb https://t.co/NXrSXPX7VX - **Aly_Cabral** (2026-07-27): https://x.com/Aly_Cabral/status/2081757877870456991 - you give feedback on the butter ratio one time.... https://t.co/UKHsmeacvW - **sarahcat21** (2026-07-27): https://x.com/sarahcat21/status/2081750257503732074 - I still remember my first Strangeloop. Meeting people from Italy to Kansas. Discussing everything from automated performance profiling to the advantages of Clojure. Listening to @emeryberger present (still among my fav talks of all time). Riding vintage slides. I want that magic back. Not just the cool after hours venue or the wonderfully weird people but also an opportunity to hear about the strange but also strangely relevant stuff that doesn’t get airtime on X. So we decided to host a conference… - **soubiran_** (2026-07-27): https://x.com/soubiran_/status/2081728246060576782 - ok so @CloudflareDev Artifacts and @flueai is insanely powerful, can't wait to share more about it! https://t.co/NvJGDsxdpu - **clairevo** (2026-07-27): https://x.com/clairevo/status/2081563367710564473 - I’m on vacation and have decided there are two kinds of people in the world: - those that smile at babies on planes, in restaurants, and hotels - and those who don’t - **ericclemmons** (2026-07-26): https://x.com/ericclemmons/status/2081446802679562348 - My current experience with AI feels like expanding from a T-shaped engineer to V-shaped. I’m afforded the opportunity to go deeper into surrounding areas than before. https://t.co/fnaUWnIQHm - **cremieuxrecueil** (2026-07-26): https://x.com/cremieuxrecueil/status/2081446343671693644 - Yes! These people have made building in San Francisco so restrictive that OVER HALF of the homes there now are illegal to build today For SF to get in compliance, 130,860 homes would have to be destroyed This self-inflicted housing shortage is the city's biggest problem—by far https://t.co/fosP33bii8 https://t.co/FB50rLop72 - **_heyglassy** (2026-07-26): https://x.com/_heyglassy/status/2081396451532509400 - TIL this is powered by PlanetScale Postgres with cf hyperdrive The list of things I use day to day powered by us is growing https://t.co/globwuOzpZ - **soubiran_** (2026-07-26): https://x.com/soubiran_/status/2081388993665503637 - sooo @flueai made its first commit on @CloudflareDev Artifacts 🤯 https://t.co/DlMdDoEUjN - **scottastevenson** (2026-07-26): https://x.com/scottastevenson/status/2081222907326304524 - I have great respect for YC, but need to say what many are surely thinking: Startup School in a stadium feels antithetical to the YC brand and to what makes startups work. The YC brand felt like it was designed to filter out status chasers. The bare Mountain View office. The utilitarian design of Hacker News. Everything about YC said: if you’re a curious hacker, we’re your people. If you’re looking for other kinds of prestige or excitement, we’re not for you. Great early stage startups are designed the same way: they’re for the people that get a thrill from the work itself, not surrounding fanfare. If it were just the arena I wouldn’t comment, but you can feel the gravity of a status vortex slowly tearing down what made YC special. You can feel the herd rushing into an overflowing barn. A stadium sized barn. You see how YC is now more prestigious than most college degrees. How prestige seekers name drop YC partners. How many YC founders now quit in the early innings—because they didn’t really love the work. Maybe it’s inevitable. It happens to startups too. Eventually the underestimated, curious, hard working teams become the winning teams. And then the status chasers flock to them. If YC is not strong enough to fend it off, maybe no one is. - **cramforce** (2026-07-26): https://x.com/cramforce/status/2081220805094936947 - Our wedding band by the way https://t.co/T0EA3LTbpG https://t.co/k0clQDSp7v - **KyleVansice** (2026-07-26): https://x.com/KyleVansice/status/2081193462108717362 - You really can’t be angry enough at what the “build nothing” crowd has taken from us. Beauty, community, prosperity, opportunity, imagination, everything. This should be Eden. But we will win, and we will build it. https://t.co/LacqYJ2DGP - **evanjconrad** (2026-07-26): https://x.com/evanjconrad/status/2081189128365015180 - when rent is $8k for a one bedroom, please remember that the fault is in the people who, for 50 years, showed up to argue against nearly every unit of housing proposed these people are no san franciscans https://t.co/QrGfIFYBnd - **steipete** (2026-07-26): https://x.com/steipete/status/2081175795587072421 - 1) Competition is good for the ecosystem. 2) Serving models at scale is hard. Proud that @OpenAI signed the letter. Ant is silent on it. https://t.co/r9QVQ2R9JH - **mrlesk** (2026-07-25): https://x.com/mrlesk/status/2081151634223026308 - Turns out your ChatGPT Sites are not just beautiful, they are quite performant too! https://t.co/sIUiqK1TZB https://t.co/SKwSnLb0yS - **grinich** (2026-07-25): https://x.com/grinich/status/2081144924112822392 - 🥰 https://t.co/G3JClWrRln - **kane** (2026-07-25): https://x.com/kane/status/2081141625611575598 - The reason new buildings in San Francisco look like “ugly plastic boxes” is because it’s mandated by @sfplanning red tape. @sfgov literally made ugly buildings the law and then let busybodies block new housing by claiming it’s ugly. @sfplanning should be abolished. https://t.co/cdBXMpx1ak https://t.co/yv4iKFAYqx - **leinwand** (2026-07-25): https://x.com/leinwand/status/2081132037147349014 - Got nerd-sniped by a single internal chat message and that was… checks watch… three hours ago. Never stop learning. Time to go touch grass. ☀️ - **chrismunns** (2026-07-25): https://x.com/chrismunns/status/2081066901820543146 - Triple the size. Build 3 day cares onsite. Build a giant kids play place. Bring down the average age of this neighborhood. Anything to shrink his boomer influence on the city. https://t.co/CZro3tqaGa - **kepano** (2026-07-25): https://x.com/kepano/status/2081059135140032915 - The extent to which Claude specifically is being promoted by bot accounts on GitHub and social platforms smells fishy. It forces me to wonder how much of it is an organized campaign and who is paying for all the tokens. - **elithrar** (2026-07-25): https://x.com/elithrar/status/2081036176111161719 - should I hook this up to wrangler? dev, dry run, deploy… what else? https://t.co/aN7fNvwVc3 - **MattieTK** (2026-07-25): https://x.com/MattieTK/status/2080985089781973074 - @thdxr attn dax fans who are getting tired of wrangler for whatever reason (we hear you) 1) tell me why (wrangler pm) 2) if you're a cf customer, reply to this post or fill out this form and I'll get in touch soon to show you something you might like more https://t.co/tAdP665RgP - **grinich** (2026-07-25): https://x.com/grinich/status/2080905267836043336 - This is now officially published on the Chrome Web Store 🎉 Try it today: https://t.co/4xL6Yrihs4 https://t.co/Pq6HZkXQl7 - **thsottiaux** (2026-07-25): https://x.com/thsottiaux/status/2080881543267172857 - @betomoedano Who is gonna tell him - **zeeg** (2026-07-25): https://x.com/zeeg/status/2080880986808816086 - we are not the same https://t.co/taRL13IPST - **thsottiaux** (2026-07-25): https://x.com/thsottiaux/status/2080869898339991732 - @theo 8 days to go - **lessin** (2026-07-25): https://x.com/lessin/status/2080829294255755476 - this is really an incredible image of who is blocking things … https://t.co/DAWWm7lE2V - **harshil1712** (2026-07-24): https://x.com/harshil1712/status/2080799754607272179 - Highlight: build time cut down from ~19 mins to ~7 mins 🔥 https://t.co/m5ofXo7gg3 - **MarkovMagnifico** (2026-07-24): https://x.com/MarkovMagnifico/status/2080795773784948839 - imagine an interest group trying to build a highway here. this is the effect the boomers have with their grip on society https://t.co/v3VY7PyFmB https://t.co/hhr0NT3MEs - **KentonVarda** (2026-07-24): https://x.com/KentonVarda/status/2080785595807228127 - I would honestly go a step further than all these people and say: Our goal should be that, eventually, all frontier models are open-weight. There may be some messiness in the short term around making sure defenders have a chance to fix security bugs before the models reach attackers. But this should be seen as temporary. We can reach a steady state where offensive capabilities of these models aren't a big deal because we've already prepared defenses. This has frankly been the story of computer security all along, it has always been an arms race, but things have nevertheless gotten broadly better over time. It'll be OK. Otherwise, the only long-term benefit of closed models seems to be that they are easier for the labs to monetize in order to fund training of new models. But do they really need that? The Chinese labs seem to have found business models that don't require this. And does charging for inference even work as a business model? Competition -- including from Chinese labs -- seems likely to drive inference profits to zero anyway. And model makers charging per token creates all sorts of perverse incentives. I think we should find some other business model here. I think we should even be willing to consider (gasp) government subsidies. AI will create value across the entire economy. What if we pay for model training by raising corp taxes, and then make all the weights open? Just throwing that out there as a straw man -- the real answer will be more complicated, I'm sure. If we *don't* do this, then frankly it seems to me that Chinese models will dominate the world, not necessarily because they are the best (though maybe they will get there), but just because they are open and ours aren't. It would be a pity if that's why American models lose in the end. - **mattzcarey** (2026-07-24): https://x.com/mattzcarey/status/2080777090190680360 - stateless mcp https://t.co/rLCLJ3lRhS - **FeifanZ** (2026-07-24): https://x.com/FeifanZ/status/2080777081580085468 - @irvinebroque i do actually think this’ll get easier as models and harnesses get better: we now need less scaffolding around models bc it’s easier to get them to do the right thing and also fix problems when they come up https://t.co/8VMSArQ9VF - **minglu** (2026-07-24): https://x.com/minglu/status/2080754890759176592 - now that we've settled which gateway is fastest... you can also run Opus 5 now through Cloudflare's AI Gateway: https://t.co/QolmZixIOl - **BarackObama** (2026-07-24): https://x.com/BarackObama/status/2080752605278449726 - Got the chance to see these talented players in action today! It was great having the @WNBA at Home Court at the Obama Presidential Center for All-Star Weekend. https://t.co/3UBiu9HS4K - **villi** (2026-07-24): https://x.com/villi/status/2080744401379397694 - Look at the average age in the room. All of these boomers live in homes that are worth several million dollars, and they gathered to protest building housing so that young people can buy a home. https://t.co/Lc4YPFvetj - **ModernNotoriety** (2026-07-24): https://x.com/ModernNotoriety/status/2080733845108699545 - Paige Bueckers and Kevin Durant hooping in Chicago 🏀 https://t.co/aIDS9X04Gh - **thdxr** (2026-07-24): https://x.com/thdxr/status/2080723541490573601 - cloudflare access has been very useful for this era of small vibecoded tools we often have something that needs to be public facing but then we configure /admin/* to need google login it's part of our IaC setup so the agent can both create this ui and protect it - **momito** (2026-07-24): https://x.com/momito/status/2080716628216397889 - Introducing Nimbus - docs for the agentic web 🌧️ Nimbus is an Astro framework for building docs, made for human and agentic workflows. - own every file, no theme to fork - agent-native: llms.txt, .md twins, MCP - features install as recipes your agent applies - support for agentic authoring, with a built-in prose linter - open source We've dogfooded it on our Cloudflare Docs this week (which has 8,700+ pages) and cut build times from ~19m to ~7m now. Your docs are product zero, they deserve the same care as the product itself. It's still really early and a bit of an experiment, so I'm excited to see where this ends up. - **gokulr** (2026-07-24): https://x.com/gokulr/status/2080705204677403135 - Phenomenal quote by @sherylsandberg (quoted by @boztank in his interview): Companies rarely fail by setting hard goals and missing. They fail by setting easy goals and hitting every one. - **badlogicgames** (2026-07-24): https://x.com/badlogicgames/status/2080704350339387666 - recommended reading. till the end. https://t.co/hwlJJK6fRr - **vjon** (2026-07-24): https://x.com/vjon/status/2080687945762349516 - Hundreds of people with $3-5 million homes in the second most expensive city in the hemisphere gather to put a stop to “greedy developers” building a tower with hundreds of $1.5 million “luxury condos” that will each pay 5x more property tax than them. https://t.co/qrO9r4p23O - **thdxr** (2026-07-24): https://x.com/thdxr/status/2080678918277771630 - if you watch just one part of this watch the "16 year overnight success" section https://t.co/2hmSY7m5yG - **michellechen** (2026-07-24): https://x.com/michellechen/status/2080677680425345316 - day 2. still on top. landed a bunch of perf improvements. we love a goalpost and rallying behind a mission. enjoy! https://t.co/YQMRciPI0v https://t.co/AlTATDwAsO - **FredKSchott** (2026-07-24): https://x.com/FredKSchott/status/2080670327903052046 - how far can we take these composable agents? here's MCP support with dynamic auth: https://t.co/FKCJ8l4V3W - **KentonVarda** (2026-07-24): https://x.com/KentonVarda/status/2080666760584241599 - Building software by prompting agents without ever reading or editing the code directly isn't actually a new thing. It's just called being an engineering *manager*. We have decades of experience about how to be effective as engineering managers. It is a different skill, though. - **KentonVarda** (2026-07-24): https://x.com/KentonVarda/status/2080666762140402141 - Don't get me wrong, I have spent my whole career avoiding being a manager and I'm not eager to be one now. But it's funny to see people talking about ensuring AI code quality by having AI review other AI's code, setting up testing regimes and loops and stuff, as if this is all new. That's what managers have been doing all along, with their human reports. You are just a manager now. - **boristane** (2026-07-24): https://x.com/boristane/status/2080523922760724886 - I just realised I own lebron dot dev - **_xjdr** (2026-07-23): https://x.com/_xjdr/status/2080401621708922888 - i wrote an essay about this a few weeks ago but was going to save releasing for a bit. seems like a good time to do it now https://t.co/ehDeY8rK2l https://t.co/acOQzL8v28 - **parkerconrad** (2026-07-23): https://x.com/parkerconrad/status/2080372277573451927 - If Anthropic is right and distillation is impossible to prevent, it follows that the strategic value of anyone opening a lead on frontier model capabilities is short-lived, and national security arguments for protectionism fall apart. If china pulls ahead we can distill them! - **QBacquele** (2026-07-23): https://x.com/QBacquele/status/2080353807179927677 - 1/ The world's songbirds sing in just 8 basic sounds. And where a bird lives decides which ones it uses. My first PhD paper is out now in @ScienceMagazine (https://t.co/Lrh8HVdIDT) Here is what we found: https://t.co/iECBEeBpPJ - **southpolesteve** (2026-07-23): https://x.com/southpolesteve/status/2080345888736571661 - Been a while since I tweeted about hiring. For the record: I'm always hiring Right now I have two high priority EM roles open. One for Agents, one for CLIs/Frameworks. Both are east coast / EU preferred. Interested? DM or email me - **meaganrgamache** (2026-07-23): https://x.com/meaganrgamache/status/2080335132515611111 - Now that the news is out — I'm hiring for R2! Reach out! - **dok2001** (2026-07-23): https://x.com/dok2001/status/2080329539885150619 - @HeyGarrison Agree. Everything I have seen from @HeyGarrison is class act. - **HeyGarrison** (2026-07-23): https://x.com/HeyGarrison/status/2080318567963066389 - "usually vendor pressure tipping" This is true. I think you would be surprised how we've withstood pressure to influence results. Every time we face these situations it becomes more clear why our independence is so important. This is also why we do all of our benchmark work in public and run on a consistent basis. - **reach_vb** (2026-07-23): https://x.com/reach_vb/status/2080316924760572055 - @simonw Sites also works in Codex mode, the only mode it doesn't work in is Chat. I'm going to make docs clearer on this: https://t.co/7OnHk2WGyQ - **simonw** (2026-07-23): https://x.com/simonw/status/2080315993101115485 - ChatGPT Sites means ChatGPT in "Work" mode can build and deploy public websites running on Cloudflare Workers, including with persistence on top of SQLite (OpenAI do not make it easy to figure out that's how the platform works, though) https://t.co/DOvq0WyTXO - **floydophone** (2026-07-23): https://x.com/floydophone/status/2080312132365701574 - hes right https://t.co/ayjnzRGPSI - **dok2001** (2026-07-23): https://x.com/dok2001/status/2080308542050083059 - The reality is most benchmarks are BS. Look behind the scenes and there's usually vendor pressure tipping the needle one way or another. Regardless, the @Cloudflare team loves a good stick in the ground to go optimize. Performance and security are our business. PS. A few more optimizations coming soon. - **unclebobmartin** (2026-07-23): https://x.com/unclebobmartin/status/2080257779395154409 - I’m significantly older than you. I started coding in the late 60s. My current strategy is to not read any of the code written by my agents. That’s the only way I can take advantage of their productivity. What I do instead is to surround the agents with extreme constraints. Unit tests, gherkin tests, QA procedures, quality metrics, mutation testing, test coverage, and a plethora of others. In the end, I have very high confidence in the code they produce because they’ve had to run the gauntlet of all of my constraints and tests. - **mitsuhiko** (2026-07-23): https://x.com/mitsuhiko/status/2080243367535898707 - We've recently made Pi's cache behavior more visible. This site has been debating whether agent harnesses are helping or quietly torching their caches. That seemed like a good excuse to explain how KV caches actually work and how Pi helps (or doesn't). https://t.co/fXwUFHLAD6 - **ritakozlov** (2026-07-23): https://x.com/ritakozlov/status/2080222204537573605 - been thinking about what makes ai writing so glaring that it’s hard to unsee even when edited down some of it is the emdashes, the lack of typos and imperfections and the repetitive longwinded nature. and the little formatting idiosyncrasies we’ve picked up “it’s not foo, it’s bar.” but i don’t think that’s all of it i review a lot of writing at work: a lot of public and internal blog posts, problem statement docs, PRDs, presentations even there’s this thing that happens, especially with junior folks. they assume that they need to sound “professional”. this is what we teach at school: start with a hook, 5 paragraph essay, use big words to sound smart, length requirements over brevity even before AI this fascinating thing would happen where i would get something to review that would sound like joey’s letter from friends. every word has been replaced with a thesaurus definition to the point of being incomprehensible the result was always jarring and unreadable at first. then i’d ask them to put down what they wrote and say it in their own words. now it would sound normal, and importantly, like them! it reminds me that we trained ai to be this way. AI is merely a reflection of what we’ve taught it and of course it’s not just junior people. if you’re not used to writing, it’s intimidating! it requires staring at a blank sheet of paper and scrutinizing words and ideas as they come to you. it’s easy to reach for the cheat code this is all to say what’s been said before: in an era of AI slop, lean into your voice and your authenticity. let it be choppy and imperfect and expressive and quirky (and personal!). i’m actually not saying “don’t use AI” to help you write. i myself reach for it quite often — it’s a helpful rubber duck (btw i typed that em dash myself ty). but don’t forget that your voice is unique to you and what makes the writing good. don’t let ai rip out the best part - **zeeg** (2026-07-23): https://x.com/zeeg/status/2080118218233114628 - mcps are much stronger typed, more native capabilities, generally should be overall better also dont forget: skills are generally token hungry, and llms are not perfect at writing bash, meaning more errors but the fact that this conversation is going makes me think anyone asking about it isnt serious like you dont gotta have a PhD to reason about how token consumption and recall work - **brendan_j_ryan** (2026-07-23): https://x.com/brendan_j_ryan/status/2080105929954271426 - Ripe for disruption!! https://t.co/mhM2jRNgij - **ericclemmons** (2026-07-23): https://x.com/ericclemmons/status/2080092039882867150 - Thanks @sdamico https://t.co/KSk9EbHxud - **jonas** (2026-07-22): https://x.com/jonas/status/2080061349246177363 - The DO can just sit on the outbound request while I get to the approve button - **jonas** (2026-07-22): https://x.com/jonas/status/2080061259718754540 - Fetching through durable objects in genera is OP. It's also how we do human in the loop approval - **jonas** (2026-07-22): https://x.com/jonas/status/2080061131368853915 - Do you want your AI agents in their little sandboxes to use your API keys without ever seeing them? Here's a cool thing you can do with cloudflare durable objects that I haven't seen elsewhere You can model the secrets as durable objects that you fetch _through_ in order to use the secret 🤯 Each secret is bound to an egress host and the only codepath that can ever read the secret is a fetch to that host And even better, you can then use the cloudflare container magic HTTPS MITM interception to let any program in a container use your secret without ever seeing it (yes yes in reality it's a bit more complex because we encrypt the secrets, allow multiple egress hosts, support websockets, etc etc) - **southpolesteve** (2026-07-22): https://x.com/southpolesteve/status/2080041765227127161 - Thank you @kuizinas - **meaganrgamache** (2026-07-22): https://x.com/meaganrgamache/status/2080039653022052487 - Some news! I’ve joined Cloudflare to lead Developer Platforms. We’re at an inflection point in our industry. AI is changing what is needed from infrastructure, and Cloudflare is uniquely positioned to meet the moment. The pace, energy, and momentum on the team right now is incredible. If you’re a developer building on our platform, I’d love to hear your feedback. If you’re looking to join a high-caliber, high-velocity team, we’re hiring! Please reach out! Big thanks to @ritakozlov and @dok2001 for the opportunity! Let’s go! - **CloudflareDev** (2026-07-22): https://x.com/CloudflareDev/status/2080034546397872400 - Devin can now run on Cloudflare using Devin Outposts. Run Devin agents in isolated sandboxes orchestrated by Workers, while controlling how they access internal services through your proxies and private connectivity. https://t.co/QuGgcfn2Wt - **Vercantez** (2026-07-22): https://x.com/Vercantez/status/2080020439196397622 - underrated feature of cloudflare ai gateway is the ability to use custom providers. we self host deep seek on AWS instances and fallback to azure and openrouter and it 'just works' https://t.co/VITeVquBoe - **southpolesteve** (2026-07-22): https://x.com/southpolesteve/status/2080008161679823127 - Zod as a compiler plugin! https://t.co/qxO7iTBHgm Works with every bundler but I'm partial to vite. Anyone tried this out? Doing schema parsing and construction at runtime has always struck me as odd. This should happen at build time. - **mitchellh** (2026-07-22): https://x.com/mitchellh/status/2079992025261424803 - Wrote up a practical introduction to SIMD using a real example from Ghostty. SIMD has a reputation for being complex, but the common case follows a simple, repeatable shape and shouldn’t be scary! Everyone should know at least this much. https://t.co/yVcAycaq8P - **FredKSchott** (2026-07-22): https://x.com/FredKSchott/status/2079980482868183368 - what if your agent could upgrade itself? https://t.co/BMt9HL89qe - **FredKSchott** (2026-07-22): https://x.com/FredKSchott/status/2079979676911714379 - in the next version of flue: composable agents define your agents with code — not config — and react to different conditions to enhance or customize agent behavior as the conversation evolves. https://t.co/yJWQS8TvJo - **FredKSchott** (2026-07-22): https://x.com/FredKSchott/status/2079930714699788789 - fine, i'll build it. https://t.co/6o7Jgov9dD - **yusukebe** (2026-07-22): https://x.com/yusukebe/status/2079825908547060206 - 個人開発Cloudflareで全部できる2026年夏版 * Webアプリ => Workers * オブジェクトストレージ => R2 * SQLデータベース => D1 * キーバリューストア => KV * セッション管理 => Durable Objects * 定期実行/cron => Cron Triggers * キュー => Queues * リトライ、スリープ処理、長時間タスク => Workflows * メール受信送信 => Email Routing / Sending* * レートリミット => Rate Limiting * AI推論 => Workers AI * ベクターデータベース => Vectorize * ヘッドレスブラウザ => Browser Run * 画像リサイズ => Images * フューチャーフラグ => Flagship * キャッシュ => Cache API / Workers Cache * バックエンドフレームワーク => Hono * フルスタック => Astro / TanStack系 / vinext ほぼ全部無料で始められる！！！！ デモ: https://t.co/wJS1RHfaoI 使って！！ - **swyx** (2026-07-22): https://x.com/swyx/status/2079775327539339329 - at some point in your engineering career, a wizened graybeard is going to lecture you about the importance of independently separable control plane vs data plane. it is very important that you listen. (and then try to learn about the management plane as early as you can..) https://t.co/EQZLBKcHof - **ericclemmons** (2026-07-22): https://x.com/ericclemmons/status/2079763976729092174 - This is the way. Ex: I’m improving Pages & Workers Builds. I’ve opened PRs or discussions for Workflows, Containers, Flagship, Wrangler, Agents SDK, Dashboard UX, CI/CD, and [redacted]. I honestly *love* being able to feel some pain and remove it for the next person. https://t.co/rH0W8cQ1wu - **thdxr** (2026-07-21): https://x.com/thdxr/status/2079692138439229758 - the agent you're building into your product is most likely not going to work just drop it, what's going to happen is there will be a few agent products that individuals and teams choose to use and they will want everything there. might mean they don't even want your product - **mattsimpsn** (2026-07-21): https://x.com/mattsimpsn/status/2079690718394605603 - been testing this out all week… it’s sooo good! fred cooked 👀 https://t.co/b29oUIAcup - **akothari** (2026-07-21): https://x.com/akothari/status/2079654012371230968 - There’s an insane amount of alpha in: - People in SF leaving SF to see how people and companies around the world are using AI (hint: most are still using it primarily as search) - People outside SF spending even two weeks in SF to see how people and companies are operating here (hint: many are already living in the future) This may be obvious to some, but the gap right now is staggering. The global diffusion of AI is a massive opportunity. - **Baconbrix** (2026-07-21): https://x.com/Baconbrix/status/2079650870254571560 - Honored to announce that I’ve joined @SpaceXAI! 🚀 I believe this team can do for intelligence what SpaceX did for rockets and the internet, I'm grateful and humbled to contribute to it. If you have feedback on Grok, Grok Build, or the APIs; I'm all ears! https://t.co/tNeJos2TDZ - **FredKSchott** (2026-07-21): https://x.com/FredKSchott/status/2079630772928147676 - is there really no "React for Agents" yet? https://t.co/MJlz1OVKJ6 - **ghostwriternr** (2026-07-21): https://x.com/ghostwriternr/status/2079624882233413871 - You can now run Devin on Cloudflare Sandboxes! Single click to set it all up on your Cloudflare account : ) https://t.co/KUC5baQ93b https://t.co/bpyg2IVMNw - **alexisohanian** (2026-07-21): https://x.com/alexisohanian/status/2079612368225755267 - The people who can hold the whole system in their head are gonna run circles around everyone else. Been true forever, just way more obvious now. https://t.co/5c5VPMPB4V - **cognition** (2026-07-21): https://x.com/cognition/status/2079612232284229952 - Run Devin Outposts on @Cloudflare Workers to give every session an isolated sandbox running on Cloudflare’s edge network, with customizable proxies and private connectivity to internal services. Deploy on Cloudflare: https://t.co/cLamv70Its - **burcs** (2026-07-21): https://x.com/burcs/status/2079572500267049076 - new cloudflare docs site just shipped and it's beautiful built using nimbus (a new documentation framework built entirely on top of astro) https://t.co/r011nADpTp - **BraydenWilmoth** (2026-07-21): https://x.com/BraydenWilmoth/status/2079571550722834465 - The entire Cloudflare Docs just got a huge glow up making it easier to read with a better navigation experience. It looks so good. Blog last week. Docs this week. What next? s/o @momito https://t.co/dtTfZZNdts https://t.co/sTYazWxtac - **thdxr** (2026-07-21): https://x.com/thdxr/status/2079561260765446587 - 5.6 is the most usable model ever made not saying it's gonna magically solve all your problems, it's just very easy to get good results from with low effort at this point if you're still struggling to get good code out of the machine it might be a you thing - **thdxr** (2026-07-21): https://x.com/thdxr/status/2079552220727562295 - it is crazy how stupid we are infinite examples of export controls being a stimulus for domestic production this isn't some hidden secret counterintuitive concept. they teach this to econ undergrads https://t.co/Lk0N8oXCBF - **c_pick** (2026-07-20): https://x.com/c_pick/status/2079236233939546281 - After using nothing but Cloudflare for a year I'm working on a client project that needs to use AWS and jesus christ I cannot believe people live like this - **omooretweets** (2026-07-20): https://x.com/omooretweets/status/2079226391594467397 - I spent most of my weekend on ChatGPT Sites. Here is why I think it’s the biggest unlock yet in the mainstream consumer being able to build with AI: - Aesthetics. Almost everything I have tried to build on Claude has a similar format, similar font, similar overall (recognizable) feel. IMO, it’s great at putting things into a template - less good at creative choices. I got very different and more opinionated styles on Sites even with very little prompting. - Hosting. If you’re a non-technical person who has tried to deploy a vibe-coded website, it’s pretty much a nightmare. Figuring out what you need to do to buy a site, handling security, setting up auth if needed, etc are all big obstacles. This is even trickier with API keys that need to be stored properly. Publishing on Sites is literally one click. - Sophistication / interactivity. I was shocked by what 5.6 could one-shot within Sites. It built fairly complex real time game mechanics (ex swipe on cards, move around in Google Street View on a real map) with extremely simple text prompts. It would also make good choices towards making those interactive experiences work - ex. a “Bonus” page in a game should pop up with a different color and UI - Identity / Auth. Sites utilize “Sign in with ChatGPT” - this is the first step in making “bring your own token budget” or “bring your own context” a reality. And yet, as the host the deepest analytics I could see was aggregate number of users and what they engaged with (no identifying info on anyone). Right now, you can set a Site to private or public - I foresee more to come with org-level controls, password protected access, etc. - **TheRealAdamG** (2026-07-20): https://x.com/TheRealAdamG/status/2079192248801988846 - https://t.co/B7MOKxtYSo ChatGPT Sites is a killer app. ChatGPT sites are a way to create and share a live website or lightweight app from ChatGPT, Work or Codex. You can learn more here: - **grinich** (2026-07-19): https://x.com/grinich/status/2078681383451627828 - weekend project: i reskinned X @chat better UX: full-screen, keyboard nav through threads better AX for my agent: it speaks WebMCP so claude can read and answer my DMs runs as a chrome extension on X's real UI so even encrypted chats work open source: https://t.co/IrzWXP1DYG - **gokulr** (2026-07-19): https://x.com/gokulr/status/2078675100367520215 - PRODUCTSPEC UPDATE: REPO MAINTENANCE We just added repo maintenance commands to the open source ProductSpec repo. (link in comments) ProductSpec started as a way to make product intent readable by humans and executable by agents. The next problem shows up once a team has more than one spec. You get a folder full of Product Specs, Agent Runs, Decision Traces, evidence links, dependencies, and revisions. At that point, the question changes. Which specs are ready for agents? Which specs are blocked? Which ones are missing evidence? Which Agent Runs actually satisfy the Product Spec they claim to implement? Which revision did the agent build against? So we added 3 commands. productspec garden: Scans the repo and reports spec health: Product Specs, Agent Runs, Decision Traces, graph health, missing evidence, stale revision pins, run gaps, unscoped specs, contention, and build waves. productspec reconcile --against : Checks whether one Agent Run satisfies one Product Spec revision before anyone claims the work is complete. It catches missing checked items, failed items, stale revisions, passed items without evidence, and drift without a Decision Trace. productspec serve --port 4317: Opens a local read-only dashboard over the same repo health report. No hosted service required. No files uploaded. Just a local view of the ProductSpec library in your repo. This matters because agent-led development needs more than a good prompt. It needs a durable product contract. It needs revision pins. It needs evidence. It needs a way to tell when the agent built what the team actually intended. The Product Spec defines the intent. The Agent Run records what happened. Related Artifacts attach evidence. Decision Trace records what changed when reality disagreed. And now garden, reconcile, and serve help teams operate that loop across an actual repo. If you're experimenting with AI coding agents in a real software repo, try running: productspec garden That is the first command I want every agent and every product team to run before picking up consequential work. - **kepano** (2026-07-18): https://x.com/kepano/status/2078596196546011254 - more on pace layers https://t.co/JAoRD8luUZ - **dexhorthy** (2026-07-18): https://x.com/dexhorthy/status/2078592010852982977 - today's fun ai coding tricking - when doing the high-level part of your plan, do an amazon-style "working backwards" approach - write the customer-facing blog post about the feature BEFORE you start building this can surface so many edge cases and help you dial in on prioritizing the important parts, and just is much nicer to read (plus you can post it after you ship 🙂) - **dexhorthy** (2026-07-18): https://x.com/dexhorthy/status/2078578258103427395 - its almost as if I wrote an entire AI Engineering guide who's core thesis was on how graphs are better than loops over a year ago enjoy - https://t.co/9iXhtq1qvd - **kepano** (2026-07-18): https://x.com/kepano/status/2078550254027477399 - formats evolve over decades, apps over months, intelligence is evolving every day in terms of pace layers "file over app" defines priorities: invest deeply in formats, and don't expect anything at the intelligence layer to last very long https://t.co/N8824IM7MA - **omooretweets** (2026-07-18): https://x.com/omooretweets/status/2078506788702629984 - ChatGPT Sites is severely underrated - it is THE breakthrough in bringing vibe coding to the masses I prompted 5.6 to make a GeoGuessr style game where you compete against LLMs Sites built the game and fully hosted it including Auth with ChatGPT, secure key storage, etc. 👇 https://t.co/JAAvnlS3cn - **thdxr** (2026-07-18): https://x.com/thdxr/status/2078496161531601060 - anyone talking about their custom setups to make ai more effective isn't doing anything useful coding used to be manual labor. typing the code out, doing large changes, jumping between files. a fancy, bespoke setup could make a difference but it's not like that anymore, you have a magic box you can say stuff into and things happen people haven't understood this, they're still spending their energy tinkering with what's behind the box. the underlying tone is always frustration. they're clearly not happy and maybe even on a burn out path the best people at this are doing the best work of their life all while happier than ever they could do this even on a vanilla setup the box can already do amazing things. it can even produce the best code of your life. it's limited entirely by the person using it and their creativity and their ambition and you're not gonna get that in a skill file - **pvncher** (2026-07-18): https://x.com/pvncher/status/2078439284407709889 - Low key you can prompt for some crazy shit when you get deep into the new multi agent architecture https://t.co/kLquFkCoNZ - **mattzcarey** (2026-07-18): https://x.com/mattzcarey/status/2078408678663700556 - dynamic workflows are graphs on demand let the agent build a control flow. https://t.co/oT8eDCTHd0 - **thsottiaux** (2026-07-18): https://x.com/thsottiaux/status/2078310751878647932 - GPT-5.6 Sol confirmed to be an extremely good model https://t.co/nZaAnFhsPX - **ianlivingstone** (2026-07-18): https://x.com/ianlivingstone/status/2078288871612186692 - Loops are inevitable - software is verifiable it’s why we have so much progress in this area. https://t.co/LkmjAsV5RF - **elithrar** (2026-07-18): https://x.com/elithrar/status/2078281560248271330 - @just_be_dev @DoorDash @acoyfellow cc @irvinebroque @KentonVarda - **grinich** (2026-07-17): https://x.com/grinich/status/2078246164550516736 - Yesterday, WorkOS experienced a service disruption that caused elevated latency, timeouts, and authentication failures for many customers. The incident was triggered by an unusual increase in session-creation traffic, but the root cause was ours: a latent transaction bug and insufficient workload isolation allowed non-critical analytics work to exhaust a shared database connection pool. We fixed the defect, restored service, and are moving analytics out of the critical authentication path, adding stronger overload protection, and improving our ability to detect and contain failures faster. Authentication infrastructure has to be dependable. We fell short of that standard and we’re sorry to our customers and their users who were affected. Full postmortem: https://t.co/qqBzezIUdI - **ChatGPT** (2026-07-17): https://x.com/ChatGPT/status/2078238669622919575 - With ChatGPT sites you can just: Build a game. Launch your small business. Make an OOO handoff. Show off your photography. Publish your research. Create a team hub. You can just build things. See the sites we’ve built and share your own. https://t.co/O1xtwuwAlz - **AdamNMayer** (2026-07-17): https://x.com/AdamNMayer/status/2078230609798045938 - I know my California urbanist/YIMBY friends don’t want to hear this: the problem is organized labor. https://t.co/0modLh8Doe - **boristane** (2026-07-17): https://x.com/boristane/status/2078218226966925685 - it's friday shipday you have code review bots, but none of them know about the production impact of your PR until today https://t.co/on6yrIlqhC https://t.co/aFFjN64rq1 - **glcst** (2026-07-16): https://x.com/glcst/status/2077894106504712539 - Almost every single time I released something in my career, I see people saying: you are clueless. You don't understand the space. You don't know what I am doing. In the early days of the Turso Cloud, there was this guy here that would every now and they remark that "Turso keeps building the wrong thing", always followed by an explanation of what SQLite *IS*, what is good at, what is not good at, how people use it, etc. Turso now has almost one signup every 5 minutes, and 700,000 databases were created in our cloud this week. On the rewrite of SQLite itself, the common accusation was that we didn't understand that people use SQLite because of its reliability, and any rewrite would fail because it had no access to their test suite. We took Turso out of beta last week and power many production users already. Before that, it was Scylla, a rewrite of Cassandra. We were roasted by many, who said that we were morons who didn't understand that Cassandra was I/O bound, so any improvements in the CPU architecture were pointless. Scylla was 10x faster than Cassandra in real life. My first experience of that was when at Red Hat we patched QEMU to be the virtualization layer for the device abstraction for the KVM Hypervisor. I was 19 at the time, and being part of the team I submitted a talk to a local conference, who rejected me saying that QEMU was an emulator, and I didn't even seen to understand the different between virtualization and emulation if I was conflating the two. The common thread in all of those is always the same: people getting dead set in their ways, always looking at the world as it is, completely unable to understand that if you just tweak things, most things can be very adaptable. Things can reach new heights. They can be used in different ways. - **fat** (2026-07-16): https://x.com/fat/status/2077817971511308449 - For the new editable diffs[dot]com functionality, we pointed agents at existing MIT-licensed editor test suites (Codemirror, Monaco, Atom, etc.) and had them analyze the nearly ~6k tests for missing behavioral coverage. After about 14 hours, we added 195 net new test cases, 30 of which exposed real oversights across selection management, whitespace collapsing, search, history, and more (we now total ~1,281 tests). - **sama** (2026-07-16): https://x.com/sama/status/2077817060068057493 - we did not have our best last 12 months ever, which is mostly my fault, but we are about to have our best 12 months to date. the team is doing amazing work and i think you’ll be very happy with what they’ve got cooking for you. i am happy about this for many reasons, but mostly because i care about our users winning. AI has to be about giving lots of people more freedom, agency, and wealth. we want to do the right thing, but we do not want to scare people into doing our thing. - **BraydenWilmoth** (2026-07-16): https://x.com/BraydenWilmoth/status/2077759155679457439 - Cloudflare Blog now has a whole new look and feel – including dark mode. Better performance, modern look, same highly technical content you expect. We built this on top of Emdash (our full-stack Typescript CMS built on Astro). https://t.co/xLCGl4OAvu s/o to the team Kody, @selfteachme, Vance, Diogo, @ascorbic, @scottbuscemi, @kblight, Donnie, @ValSocial, @burcs ++ more! - **glcst** (2026-07-16): https://x.com/glcst/status/2077759127682486561 - I am excited to announce that we are officially writing a new version of Postgres. In Rust - and creating the LLVM of databases in the process. In the span of a year, we have rewritten SQLite. Keeping the compatibility, increasing its feature set. MVCC, Types, (Live) Materialized Views, among other things. In the process of doing that, we have realized: At the end of the day, what makes SQLite special is that it compiles SQL to a database-specific bytecode. So why can't we compile *Postgres* to the same bytecode? Turns out we can. I ran an experiment called pgmicro as a way to prove this hypothesis, and it works very well. It is time to make this official, and put the weight of Turso behind it. We shall give the world a modern take on Postgres. Wire compatible, but built on a new architecture. We have already heard of others wanting to extend this. MySQL? Redis? the sky is the limit. What can we do if we do for databases what LLVM did for compilers? To prove how powerful the SQLite bytecode is, we are actually running DOOM compiled to the unmodified SQLite instruction set. And because Turso runs natively in the browser, you can play the game in your browser. With the database executing it. Read the full story below! 👇 - **glcst** (2026-07-16): https://x.com/glcst/status/2077759129041473725 - We are rewriting Postgres. And in the process, turning Turso into the LLVM of databases: https://t.co/KeoNfbzgun - **ritakozlov** (2026-07-16): https://x.com/ritakozlov/status/2077757765523493025 - cloudflare blog is glowed up — inside and out 🥹 built on emdash, ai search, the newly released workers cache, astro and hyperdrive + planetscale https://t.co/TjTXB0Ekh0 - **fayazara** (2026-07-16): https://x.com/fayazara/status/2077753436800782341 - If you need some use cases for browser run, I had built this little site. https://t.co/vOFmLQLmVM https://t.co/bx6neViWjH https://t.co/qoYuFtiRCS - **ritakozlov** (2026-07-16): https://x.com/ritakozlov/status/2077749102641639880 - btw @cloudflare product you're not using nearly enough: browser run landed just over the past few months: - webmcp support - session recordings - live view + human in the loop that allows you to step into a live session - higher concurrency limits - quick actions via workers bindings and more to come soon! the team is just on 🔥 - **steveruizok** (2026-07-16): https://x.com/steveruizok/status/2077745437415481644 - codex is really good at writing code on tldraw https://t.co/vJUuMmhmgw - **ritakozlov** (2026-07-16): https://x.com/ritakozlov/status/2077685577009267077 - last night we hosted dinner with our london @cloudflare interns i asked the interns at my table what their current favorite ai productivity hack is, and one of them said: using @supermemory i told them it was built by a former cloudflare intern and they were completely 🤯 - **thomas_ankcorn** (2026-07-16): https://x.com/thomas_ankcorn/status/2077678147290501378 - Onboarding Workers Observability onto flagship today! Building Cloudflare on Cloudflare never looked so good https://t.co/eWyLkaqbQ2 - **thsottiaux** (2026-07-16): https://x.com/thsottiaux/status/2077632589498913087 - We've had no 5h limit in Codex plus and pro for a few days. Do you think it is better or are you finding it difficult to manage the usage included in the weekly limit effectively? If we were to make this different, what should it look like in an ideal world? - **samlambert** (2026-07-16): https://x.com/samlambert/status/2077553744456437907 - i’m triggered by the way australians pronounce the word cache - **tbpn** (2026-07-15): https://x.com/tbpn/status/2077509235936821598 - Free business idea that we actually think someone should start: The Laptop Detailing Company of San Francisco Starting at $9.99 per device, a crew of expert detailers comes on site and cleans employee laptops on a regular schedule. Company would print. https://t.co/Tqz1vzKPZH - **southpolesteve** (2026-07-15): https://x.com/southpolesteve/status/2077394873901019646 - I keep trying memory tools. I keep ripping them out for some flavor of: echo "write memories to MEMORIES.md and other important info to markdown files as needed" >> AGENTS.md https://t.co/wk1Eylh89P - **zeke** (2026-07-15): https://x.com/zeke/status/2077252641646842259 - I made an OBS plugin that mimics that sweet sweet zoom effect from the @screenstudio Mac app, but instead of doing it as a post-video effect, you do it live. Tap a hotkey to zoom in on the cursor of the screen you're recording. Tap again to zoom back out. Here's the repo: https://t.co/ruF7kSSQZF - **gdb** (2026-07-15): https://x.com/gdb/status/2077202646159802492 - with chatgpt work & sol, i'm finding it incredibly joyful to just ask any question about the business and have it be thoroughly researched and answered. realizing i have so many questions i wouldn't have bothered asking because they would be too burdensome to answer. - **_dylanga** (2026-07-14): https://x.com/_dylanga/status/2077146571372716289 - Ramp Inspect hit its 1,000,000th session this week 🚀 https://t.co/bLmIqmmgTS - **matt_slotnick** (2026-07-14): https://x.com/matt_slotnick/status/2077104905656144246 - la cabra would crush it in SF - **zebassembly** (2026-07-14): https://x.com/zebassembly/status/2077076907569426723 - One of the nice parts of the LLM era is that when evaluating if I should write this bit of code in rust with wasm or just pure JS I was able to write a JS implementation and then tell it to port to Rust, and then to wasm. Wasm is ~4x faster than the JS implementation - **sama** (2026-07-14): https://x.com/sama/status/2077053226080436235 - also, a reason to favor open-source harnesses. - **harshil1712** (2026-07-14): https://x.com/harshil1712/status/2076944028994592935 - Here's a short article on how I built this Graffiti Wall: https://t.co/V9sWAzsHXA https://t.co/MkwTRlVHz3 - **theo** (2026-07-14): https://x.com/theo/status/2076898111268753898 - I think the web dev world is adjusting to AI fastest because we're used to everything changing every few weeks - **michellechen** (2026-07-13): https://x.com/michellechen/status/2076811954639778025 - we added more capacity to GLM5.2 on workers ai. faster tps and less oocs https://t.co/t6pnJDK7nj https://t.co/6Ex52tMqds - **elithrar** (2026-07-13): https://x.com/elithrar/status/2076807033299337456 - very good. https://t.co/lrZk7aZgnE - **gdb** (2026-07-13): https://x.com/gdb/status/2076797035945308364 - Try out ChatGPT Sites, makes it way more fun to communicate things within a company: https://t.co/HqBAYk9qgJ - **sama** (2026-07-13): https://x.com/sama/status/2076780425280954658 - come for the best model, stay because we don’t treat you with contempt - **citrini** (2026-07-13): https://x.com/citrini/status/2076763278491988207 - Monday - war Tuesday - hot cpi Wednesday - warsh apocalypse Thursday - war over Friday - ai bull market https://t.co/OqkbamJZPl - **justinstorre** (2026-07-13): https://x.com/justinstorre/status/2076743122231713936 - https://t.co/FBOBxnSz5S - **mattzcarey** (2026-07-13): https://x.com/mattzcarey/status/2076738812500221999 - First class MCP elicitations support is in agents@latest! Both url and form modes :) https://t.co/dBgHJmVum4 - **astrodotbuild** (2026-07-13): https://x.com/astrodotbuild/status/2076720512768594323 - Built with Astro 🚀 https://t.co/Uz2LSDhTZA https://t.co/Lbzl2orAzh - **haydenbleasel** (2026-07-13): https://x.com/haydenbleasel/status/2076714329404268576 - Introducing Blume 🪷 A world-class docs framework for everything you ship. Drop Markdown into a folder and ship a full docs site with no app boilerplate. → zero-config setup → automatic SEO and AEO → 30+ components → powered by astro + vite → open source, free forever https://t.co/fQSwLd9BM9 - **brian_scanlan** (2026-07-13): https://x.com/brian_scanlan/status/2076713135700951343 - If you are interested in pointing your agent at a small, edited, curated set of Claude Code skills/hooks that are heavily used by hundreds of engineers at Fin, we recently published some. Link below. - **jxnlco** (2026-07-13): https://x.com/jxnlco/status/2076511474600857950 - ChatGPT Sites is now in public beta. You can turn a prompt, file, or rough idea into a dashboard, project tracker, report, prototype, or lightweight app: build and edit directly in ChatGPT Work or Codex test in a private preview publish and share with a URL Rolling out across paid plans. Enterprise admins can control public publishing. - **schrockn** (2026-07-13): https://x.com/schrockn/status/2076488446961709218 - I've seen enough: After a weekend of working with Sol and Fable it's pretty clear to me that OpenAI has taken the lead on coding models. Congrats to the team. You cooked. Looking forward to Anthropic's next move! - **keyanzhang** (2026-07-13): https://x.com/keyanzhang/status/2076461227661054015 - some tips on 5.6 sol: 1. remove old slop. try disabling certain community skills and plugins, especially bundles with 20+ skills. think of 5.6 sol like someone who just grew from senior to staff or senior staff level: prescriptive guidance that used to help becomes micromanagement and makes the work worse. you can always add back what clearly helps. 2. turn on codex memory in settings. give codex feedback on what you like and don’t like, and ask it to remember. 3. you probably don’t need ultra. i do 95% of my work on sol high, sometimes use sol extra high, and have only used sol ultra for a handful of sessions. start with high and only move *up* when you’re not happy with the result. - **thsottiaux** (2026-07-13): https://x.com/thsottiaux/status/2076459871021736245 - Rest assured that GPT 5.6 Sol will stay in the ChatGPT subscription you pay for. Including Go, Plus and Pro subscriptions. At least until we ship an even better model. - **dexhorthy** (2026-07-12): https://x.com/dexhorthy/status/2076381087203475772 - i know i've been griping about a couple SOL things but overall the GPT 5.6 models are *incredible* - here's my highlights Sol - doesn't talk like a robot anymore - super well done and impressive. not quite fable-level but soooo much better than 5.5 Terra - truly a sonnet-class model, great for research subagents, and really strong for general knowledge work - nice! Sol - still not great at UI - much better, but not reaching the opus-level of "yolo in a prompt and get something beautiful without even trying" The other thing to watch out for - these models LOVE subagents and skills - very well tuned to use them, which means we had to refine a lot of things in our general workflow 1) many skills and prompts steered pretty hard to subagents "use them", "use them in parallel", etc - this is no longer necessary 2) many skills advertised hard in description field - "pick me", "you MUST use this skill whenever ..." - that is complete overkill now 3) in fact we're disabling model invocation on a bunch of skills cause I had never need a model reach for them before and now its regularly invoking skills/commands I forgot even existed 4) i had built an intuition around typing "use a subagent"-ish things when prompting. no longer necessary. instead i am building intuition around "no subagents" and i send this probably 1 in 4 prompts to Sol all around great release - reminder that ever new generation of models means you should review and potential throw out a bunch of your skills/prompts/instructions that are no longer necessary more updates as we go, what did I miss? - **satyanadella** (2026-07-12): https://x.com/satyanadella/status/2076323181154230284 - https://t.co/xv6csf1SbV - **boristane** (2026-07-12): https://x.com/boristane/status/2076317707268714619 - wish chatgpt was on durable objects to automatically sync chats between clients - **thomas_ankcorn** (2026-07-12): https://x.com/thomas_ankcorn/status/2076316042444333241 - @dillon_mulroy https://t.co/TyNpGd2wHB - **thomas_ankcorn** (2026-07-12): https://x.com/thomas_ankcorn/status/2076311215022772231 - https://t.co/pqUa7hrbsG LGTM wen 😸 https://t.co/bYWfzoytx2 - **thomas_ankcorn** (2026-07-12): https://x.com/thomas_ankcorn/status/2076297689327690199 - Open telemetry support for OpenAI's agents sdk 🤓 https://t.co/7PfLo8rSQD - **kylegalbraith** (2026-07-11): https://x.com/kylegalbraith/status/2075996720341610549 - got it fully deployed to Cloudflare directly from Depot CI and have sol working on provisioning workos. sol has done it all. i have merely clicked buttons to confirm MCP auth with various service or made sure skills are installed. fucking wild. https://t.co/X8BofPmtxy - **sama** (2026-07-11): https://x.com/sama/status/2075983427019612242 - there are a lot of benchmarks that suggest 5.6 sol is the best model in the world right now, but the most reliable way to tell is that elon is obsessed with me again - **sama** (2026-07-11): https://x.com/sama/status/2075982820322025788 - 30% of the cost was on fable at these levels of usage? https://t.co/KzFoHj7cB4 - **sama** (2026-07-11): https://x.com/sama/status/2075982617976230043 - homeboy you're the one sellling public market investors on short-term space datacenters https://t.co/wTYSA4q3Qx - **thdxr** (2026-07-11): https://x.com/thdxr/status/2075963734745342361 - our team's first week of having both sol and fable the crazy thing is 30% of our cost was on fable https://t.co/8yVddGEuBg - **thdxr** (2026-07-11): https://x.com/thdxr/status/2075810189039538352 - being in the loop is important but i keep seeing people misunderstand this - your job isn't to micromanage the agent and get it to execute its steps in a way you think is better than what it'd do on its own it's to come up with cool things for it to work on - **pvncher** (2026-07-10): https://x.com/pvncher/status/2075663535544135935 - With 5.6 Sol, a lot of people are still prompting the model exactly as they did 5.5 It's important to note that 5.6 Sol is a lot more tenacious and thorough than previously models. Check out the guide I wrote here to get better outcomes https://t.co/nk1rHm0sWY - **RhysSullivan** (2026-07-10): https://x.com/RhysSullivan/status/2075642268082741748 - It’s absurd how good OpenAI’s communication game is https://t.co/CN2bCPlDU4 - **cryps1s** (2026-07-10): https://x.com/cryps1s/status/2075639162120900766 - The most cyber-capable models should come with the strongest account protections. Starting September 1, individual members of Trusted Access for Cyber will need to enable Advanced Account Security (AAS) with hardware security keys to retain access. Those who don’t will return to default access - but they can regain access at any time once enrolled in AAS. Hardware security keys are among the strongest defenses against phishing and account takeover. AAS with TAC will support major, certified FIDO2 keys - and we are also exploring non-exportable passkeys backed by hardware. Exportable software passkeys will not be supported. You can bring your own hardware keys, or you can buy YubiKeys directly through us (https://t.co/BSR7ONKXif) with preferred pricing. Ultimately, these changes will help maximally protect accounts with access to our most permissive and cyber-capable models. - **KentonVarda** (2026-07-10): https://x.com/KentonVarda/status/2075623669171102104 - Feeling the pain a bit of the deluge of AI security reports on Cap'n Proto. 10 bugs reported in the last three months, which is more than the entire 13-year history of the project before that. None that are extremely critical (I think) but there are some near misses. Be sure to upgrade to release 1.5 (or to tip of the v2 branch) if you use it. Don't get me wrong, I'd much rather have bugs reported and fixed than not reported. But it's a lot of work. Although the Cloudflare Workers team heavily uses and contributes to Cap'n Proto, I still technically own it as a personal project and have been personally responsible for releases and security advisories. That used to work perfectly fine, I spent like two days a year on it... but I think it might not be viable anymore. It's a lot of work to write up every bug, request a CVE, etc. Rules say you must issue a separate advisory and get a separate CVE for every bug, but I just don't have time, so I issued a rollup advisory with no CVEs. Sorry. Gonna have to figure out something new going forward. Ideally the whole process could of course be automated with AI, but then I have to spend time setting up that automation, including prompting it to write advisories in a way that doesn't suck, which I... also don't have time to do right now, too much going on. https://t.co/5Wc23RacE8 - **jack** (2026-07-10): https://x.com/jack/status/2075604612359315647 - i’ve shifted from telling agents what to do, to asking them what to do, and pulling the best thread. - **ackriv** (2026-07-10): https://x.com/ackriv/status/2075593741092405543 - Smart Tiered Cache quality of life improvement just landed for those hosted on public clouds https://t.co/l2BLT8hvLh - **gdb** (2026-07-10): https://x.com/gdb/status/2075592729736995209 - I am deeply grateful for all Fidji has done for OpenAI and to advance our mission, and for the opportunity to have worked alongside her for the past few years. It’s quite sad, but I’m glad that she’s taking the time to work as hard on her health as I’ve seen her do professionally — hoping for a quick recovery. - **keyanzhang** (2026-07-10): https://x.com/keyanzhang/status/2075374891558425037 - 3D pelican riding a tricycle, plus a pelican riding a pelican, demoed by @edbayes https://t.co/vz7MwjyzdQ fun fact: no one read or edited the code, and no assets were uploaded. this was built entirely by giving 5.6-sol high simple text instructions like “set a goal to render this as accurately as you can”, “add a g-wagon”, and “can you add another ride: a pelican riding a pelican?” - **sama** (2026-07-09): https://x.com/sama/status/2075354679031067058 - i am really sad about this and very grateful for all fidji has done for openai, and even grateful for her friendship and who she is as a person. we all wish her the best for a speedy recovery. this sucks. https://t.co/tg33Lutsn7 - **fidjissimo** (2026-07-09): https://x.com/fidjissimo/status/2075353170927304861 - Today, I shared with the OpenAI team that I have decided to leave my full-time role at OpenAI and transition to being a part-time advisor. Three months ago, I had to go on medical leave after a severe exacerbation of a chronic illness I’ve lived with for seven years. During that time, it became clear that the road to recovery would be much longer and more complex than I had anticipated—and that I needed to focus on it fully. When I went on leave, many people told me I was courageous for prioritizing my health. The truth is that I am only making this decision now because I failed to make it many times before. Over the years, doctors, friends, colleagues, and loved ones encouraged me to slow down. Two years after I got sick, Facebook offered me the opportunity to take a full year of medical leave. I didn’t even pause to consider it. I immediately said no. At the time, Zuck told me I should play the long game. I wish I had listened. Looking back, I realize that a lot of what made me successful also made this decision incredibly difficult. I grew up believing that opportunities were precious and that when they appeared, you grabbed them with both hands. That mindset carried me from a small town in southern France to opportunities I never could have imagined. By the time I turned 40, I had already gotten to do more than I’d ever dreamed possible as a kid growing up in Sète. I love building. My work has always given me a deep sense of purpose. OpenAI in particular felt like a role that my entire career had been building toward, which made this decision even harder. But what I’m learning now is that grit and endurance are not the only skills required to have impact over decades. Sometimes the harder thing is to stop, listen, and trust that taking care of yourself today makes it possible to contribute for much longer tomorrow. This experience has also strengthened my conviction about why this work matters. It has been a jarring experience to spend my days helping build the future while simultaneously navigating a disabling disease that still has no cure. Over the last seven years, I’ve spent countless hours in doctors’ offices, dealing with symptoms, treatments, insurance, uncertainty, and all the invisible work that comes with being a patient. Like millions of others living with chronic illness, I’ve experienced firsthand how difficult healthcare can be to navigate, even when you have every possible advantage. More than ever, I believe that some of the most important opportunities for AI lie in helping people solve real problems in their daily lives: their health, their finances, their time and the everyday burdens that shape human experience. In particular, curing disease is the most important thing AI could accomplish. I’m excited to continue working towards cures through OpenAI but also through my work with @ChronicleBioAI and @CODA_research. I’m deeply grateful to @sama, @gdb and the OpenAI board for their support during this time and for offering a way for me to continue contributing to the mission without sacrificing my chances of recovery. I’m also so thankful to my team and the many extraordinary colleagues I’ve had the privilege to build alongside. For now, my focus is recovery. But my belief in the potential of technology to solve deeply human problems has never been stronger. - **mattzcarey** (2026-07-09): https://x.com/mattzcarey/status/2075310974047854909 - vinext :) https://t.co/zgXuqCFcKh - **EmDashCMS** (2026-07-09): https://x.com/EmDashCMS/status/2075303797845544979 - With EmDash 1.0 around the corner, we've scheduled our second Community Call! 🎙️ Join us to discuss code contributions, the roadmap, and how to spread the word about EmDash. 🗓️ 14 July 2026, at 3PM UTC 🔗 https://t.co/8bmpAoUq82 - **boristane** (2026-07-09): https://x.com/boristane/status/2075286428121264397 - claude is becoming bloated piece of crap it used to be simple: open, type, chat now there are 6000 buttons I literally do not care about - **JonathanAbrams** (2026-07-09): https://x.com/JonathanAbrams/status/2075275500092809324 - I'm really happy to announce the launch of ChatGPT Sites to all paid subscribers! Here's a site I made last night to celebrate the release of not just Sites, but also 5.6 Sol + Terra + Luna: https://t.co/wolqtRhrE6 What will you build and share? - **mitsuhiko** (2026-07-09): https://x.com/mitsuhiko/status/2075245365301850281 - As someone who maybe is not the biggest fan of Bun, I think what @jarredsumner is doing is pretty fucking awesome and I can confirm that he's great to hang out with and just very friendly. The responses to his blog post by some people are uncalled for. - **bscholl** (2026-07-09): https://x.com/bscholl/status/2075191913117012250 - Regulatory capture is evil, self-driving cars edition https://t.co/HlOSLh5i7H - **hyfdev** (2026-07-09): https://x.com/hyfdev/status/2075081740595642779 - Vite is now well past the 20k issue mark, yet #5142 — an almost five-year-old issue about chunks running in the wrong order — is still open. I'm happy to share that we're finally making progress on it. Preserving execution order across ESM chunks has long been one of Rollup's hard problems. With Rolldown, we now have the foundation to tackle it properly. Today, you can already manually enable Rolldown's strictExecutionOrder in Vite as a workaround. But we're pushing this further: redesigning the implementation around zero-cost abstractions — or more precisely, paying only for the ordering guarantees your bundle actually needs. The goal is simple: eventually, this should just work automatically. https://t.co/XQZmOzkj6l - **sama** (2026-07-09): https://x.com/sama/status/2075047983314571764 - 🫶 https://t.co/Y19TWI4HMf - **michellechen** (2026-07-08): https://x.com/michellechen/status/2074966579851194585 - big day for the homies. super excited to work with @vikhyatk and @jason_allen to bring the long awaited Moondream 3.1 to Workers AI. workers ai runs GPUs on cloudflare's edge, which is a natural infrastructure fit for real-time use cases like vision. moondream requests return in 160ms-700ms e2e. so fun to work with a team i've admired for so long. they even tuned specific kernels to fit our latency-tier GPUs. and we run this within a cog container, where we can leverage moondream's high-performance inference engine: photon. technical blog to come soon! - **mweinbach** (2026-07-08): https://x.com/mweinbach/status/2074944427689202158 - I had a pretty similar experience A bunch of my tweets about missing Fable were actually about missing GPT 5.6 Sol, but I couldn’t say it so said Fable instead Fable is really good and maybe better in some ways, but I vastly prefer GPT 5.6 Sol https://t.co/fb4hllci9w - **zeeg** (2026-07-08): https://x.com/zeeg/status/2074935120432087286 - I'm not sure there's a product in the world that gets auth right (us included) - **jayair** (2026-07-08): https://x.com/jayair/status/2074929399842078839 - We've usually stayed away from model comparisons but 5.6 vs Fable is a unique situation We've never had a case where the team is so completely convinced on which one is better Here's the timeline of our experience with it - We test early versions of 5.6 for a couple of weeks and have a great time, it feels like a step change improvement, enabling new workflows - We get to try Fable and don't think it's not as good, I personally would take this experience with a grain of salt, there tends to be a bias when trying a new model when you already like another - Fable and 5.6 are taken away because of the regulatory issues - Our team is literally depressed that 5.6 is gone, we are looking for anything that could even partly replace it - Fable comes back, and here's where it gets interesting, you would think Fable would be enough, but no, the team is still depressed that 5.6 isn't available - Then 5.6 comes back and it's immediately clear that it's just way better than Fable This situation was unique in that it was the closest we've ever gotten to having an unbiased comparison of two models - **juberti** (2026-07-08): https://x.com/juberti/status/2074906710024892694 - Today, we're launching our third-gen voice model and architecture, GPT-Live. GPT-Live is a full-duplex model with built-in async delegation, which allows it to deliver incredibly natural conversation along with the intelligence you expect from ChatGPT. https://t.co/6P71MwTfUa - **ritakozlov** (2026-07-08): https://x.com/ritakozlov/status/2074897995225575718 - happy drop day 🎤 a place for all your agent generated md -> html sites no account needed, just drop & go. and claim if you want or let it expire https://t.co/JA41XcYhTO https://t.co/QFV7pTtMIG - **BraydenWilmoth** (2026-07-08): https://x.com/BraydenWilmoth/status/2074894829616509358 - Introducing Cloudflare Drop Drop your folder in the browser and deploy it instantly on Cloudflare. Your website... milliseconds away from users on region: earth No account needed. Deployment is active for 60 minutes, then expires unless you claim it. https://t.co/Dn6b1mggqs https://t.co/FgNhjHxXzO - **chythram1** (2026-07-08): https://x.com/chythram1/status/2074892693965074844 - super excited to announce the launch of Cloudflare Drop! just drag and drop a folder or zip file with static assets and get a live URL that you can share! best part: you don't need a CF account to get started!😁 try it out here at https://t.co/4Ud11YQ65C https://t.co/U6hsEIQWet - **eastdakota** (2026-07-08): https://x.com/eastdakota/status/2074878109405306991 - Wonder what we could be building this for…? 🤔😏 https://t.co/jV5mfGJasO - **matt_slotnick** (2026-07-08): https://x.com/matt_slotnick/status/2074871197335278074 - labs pretty much only used to do technical marketing to very online people. now they're forced to market to large enterprise buyers and the broader public who are, as they say on here, turbonormies... you're going to see a lot more of these reactions https://t.co/wcuCliEEyu - **jonas** (2026-07-08): https://x.com/jonas/status/2074845346824323248 - Haven’t we all… 😭 https://t.co/KlIB5tvb7s https://t.co/e9sZYWVqjK - **Cloudflare** (2026-07-08): https://x.com/Cloudflare/status/2074842199104135561 - Cloudflare Research is building a global consensus service called Meerkat that uses a new consensus algorithm called QuePaxa. Find out more: https://t.co/UyIv5WN0Fb - **theo** (2026-07-08): https://x.com/theo/status/2074713090978132362 - I would like to add that OpenAI has been awesome to work with despite the obvious government-related setbacks. They've been more transparent than I'd ever have expected, and they did a good job keeping us in the loop during the confusion. The only "restriction" is that they've requested is that we wait on posting our formal content about the model (blog posts, videos, podcasts etc) until Thursday. They've given us ZERO restrictions on what we can and can't say. Every employee involved with early access program deserves a promotion and a shitload of praise. They're so, idk, real? Hard to find the right words. I'm just thankful they make this so easy for us - **sama** (2026-07-08): https://x.com/sama/status/2074709023807664454 - GPT-5.6 sol launches thursday! happy building - **acoyfellow** (2026-07-07): https://x.com/acoyfellow/status/2074441316902068425 - How we lookin'? Repo: https://t.co/HekQGQdAeo Docs: https://t.co/s2POADUQpC https://t.co/LYXwvc2lT2 https://t.co/WdRMy0GBpk - **nurodev** (2026-07-07): https://x.com/nurodev/status/2074433137778508144 - @irvinebroque @sadsciencee Exactly! Some kind of wrapper, or allowing direct inline symbol assignment (Though that's a bit ugly / complicated imo) https://t.co/JHcYeMuvXv - **matt_slotnick** (2026-07-07): https://x.com/matt_slotnick/status/2074307560103190835 - rather than everyone vibe coding a full stack app, the community collectively vibe codes the core 70%, leaving specialization to happen on the deployment-specific 30% - **matt_slotnick** (2026-07-07): https://x.com/matt_slotnick/status/2074307124554051925 - beginning to think there's a world in which core SaaS goes open source. crazy, i know, but... with frontier coding models, you can imagine a world with community maintained 70% versions of core SaaS, built to be configured for last mile. this didn't sense pre-AI, but might soon - **hichaelmart** (2026-07-07): https://x.com/hichaelmart/status/2074289745128194212 - @irvinebroque For anyone wondering "well what's the difference between dependabot" - the main one is an agent can tell whether an upstream update is actually applicable to you. For many, they won't be - even bug/security fixes - because you just don't use that part of the library. - **hichaelmart** (2026-07-07): https://x.com/hichaelmart/status/2074289098999894361 - @irvinebroque I haven't – so far I tend to just subscribe to GH project releases for ones I care about, but that's not sustainable. You'd want a daily cron agent to check (and I'd only care about bug+security patches tbh). Should be an easy task, but haven't tried it 😉 - **nurodev** (2026-07-07): https://x.com/nurodev/status/2074282432749007348 - @sadsciencee @irvinebroque Decorators I’m intrigued to try and see what, if any, foot guns could crop up from it. Same with a wrapper API. Though I’m also curious about trying a symbol approach. Same with a Symbol.dispose to cleanup resources out of scope, we could tie metadata to RPC results 🤔 - **hichaelmart** (2026-07-06): https://x.com/hichaelmart/status/2074282224627912878 - A risk-free, maintainable and agent-friendly future is absorbing deps into your codebase. Strip unused sub-modules, use your formatter, your bundler. Adhere to license and attribution ofc, but it's your responsibility now. Watch for upstream patches and merge in. - **samlambert** (2026-07-06): https://x.com/samlambert/status/2074273840813990104 - We had a talented engineer apply to PlanetScale but they told us they are ethically opposed to using AI. How do people think this is a possible stance in 2026? - **boristane** (2026-07-06): https://x.com/boristane/status/2074272551958966756 - sandbox founders are the new database founders - **AdamRackis** (2026-07-06): https://x.com/AdamRackis/status/2074270802241859612 - Whoa. Very cool! https://t.co/8SQOsPncoc - **warriors** (2026-07-06): https://x.com/warriors/status/2074239144143081712 - Al Horford 🤝 Dub Nation https://t.co/mR6THGwJf5 - **samlambert** (2026-07-06): https://x.com/samlambert/status/2074208900224864446 - SF now has a monthly database meetup! https://t.co/aX1YlX2clk - **grinich** (2026-07-06): https://x.com/grinich/status/2074179179646730479 - Few people have shaped the trajectory of AI more than @drfeifei, from foundational research to building at the frontier with @theworldlabs. On July 22, @dwarkesh_sp will host an off-the-record conversation on what's next. Space is limited. RSVP below👇 https://t.co/6l5gMs5nRg - **HotAisle** (2026-07-06): https://x.com/HotAisle/status/2074164437364449784 - Wow. I used to do so many hacks to get this functionality. I once built a cf worker caching layer in front of github so that I could have 30k servers downloading private repo binaries without getting rate limited by GH. Eventually hit one of cf’s undocumented rate limits and had to get an account exec to fix it. - **wongmjane** (2026-07-06): https://x.com/wongmjane/status/2074164112859549713 - @Cloudflare Yes finally!!! - **thejessezhang** (2026-07-06): https://x.com/thejessezhang/status/2074154325933424861 - https://t.co/N2gSVInZVT - **ApplyWiseAi** (2026-07-06): https://x.com/ApplyWiseAi/status/2074153996051206333 - @irvinebroque finally a real answer to the dedupe-at-the-edge problem without yeeting everything into KV. curious what the eviction story looks like once traffic spikes - **svensauleau** (2026-07-06): https://x.com/svensauleau/status/2074149964658217336 - finally possible! https://t.co/3BnPqB1ksK has been running a worker behind cache for years https://t.co/hrYn3FqBss - **itsJaimeMedina** (2026-07-06): https://x.com/itsJaimeMedina/status/2074138210591199333 - New; Cloudflare Dynamic Worker APIs now can act more like CDN assets. Instead of only caching images/CSS/static HTML well, you can cache API responses, rendered HTML, search results, product/category pages, expensive database reads, AI responses, computed JSON, etc. You can put the Worker near the database, not necessarily near every user. This is probably the most important architecture change. If your database is in one region, running DB-heavy Worker code near that database avoids slow DB round trips. Then Workers Cache can serve the result close to users. That means: app logic near data, cached output near people. - **BraydenWilmoth** (2026-07-06): https://x.com/BraydenWilmoth/status/2074138048451825898 - https://t.co/TXZhXy3dZ1 - **pedrofracassi** (2026-07-06): https://x.com/pedrofracassi/status/2074133167686975549 - the way I was literally jerry rigging something like this with Cache API like 3 days ago https://t.co/pLjIxXYRXg - **realalexniebuhr** (2026-07-06): https://x.com/realalexniebuhr/status/2074131538598351108 - One of the best updates to Workers. Have already tried it with Astro and it performs exceptionally well. https://t.co/P2wIdJFKMw - **southpolesteve** (2026-07-06): https://x.com/southpolesteve/status/2074125467255230487 - @threepointone @icanvardar Shipped https://t.co/IPqvNCK8TK - **zebassembly** (2026-07-06): https://x.com/zebassembly/status/2074123775663132917 - This is amazing. I knew of this project but I hadn't had time to actually look closely at it, this is everything I wanted for cache-infront-of-workers and more. Team absolutely cooked https://t.co/YV7V0XzTmb - **mattzcarey** (2026-07-06): https://x.com/mattzcarey/status/2074104164674650456 - proud of my last 6 months https://t.co/OJWsDVXFx5 - **dexhorthy** (2026-07-06): https://x.com/dexhorthy/status/2074037838941933632 - it feels like codex-5.5 xhigh gets ~twice as much work done in 150k tokens as fable 5 high does bunch of tasks on the flight today... got used to, with fable, we're just getting started and we're already well over 120k have several similar codex sessions on same work or similar tasks and they all have done a tone of meaningful work by the 120k point - **thsottiaux** (2026-07-06): https://x.com/thsottiaux/status/2073933490513752151 - @haider1 Ultra will be in codex. - **LambrosPetrou** (2026-07-05): https://x.com/LambrosPetrou/status/2073906630995943607 - Spent an hour today designing a feature with Fable/Opus. Then, once I saw the code and played with it, I actually changed the approach entirely. You still have to have a good feel of how something fits in the overall picture. The models are not there yet. - **plannotator** (2026-07-05): https://x.com/plannotator/status/2073845811847512372 - Plannotator 0.22.0 is out. Code Review (is not dead yet): - Guided reviews give you a list of changes with semantic overviews, ordered by importance & where you need to spend your time reviewing. Inspired by @linear, this works for local and PR changes . You don't need the best models or high reasoning levels to use this! - Pi and Copilot are added as agent review providers. - A new default `git status` view is added. - A new `commits` view is added to easily navigate all commits and view their descriptions+changes. - **boristane** (2026-07-05): https://x.com/boristane/status/2073820562028446088 - I'm betting my company on proactive agents agents that sift through 100x more data than anyone, decide what's the most critical thing to work on, and just do it I wrote down my thoughts https://t.co/5WkVlwtKwR - **acoyfellow** (2026-07-05): https://x.com/acoyfellow/status/2073814578362720562 - @irvinebroque On it - **zeeg** (2026-07-05): https://x.com/zeeg/status/2073810914902729192 - the worst part about this shift in technology: you need large amounts of automated verification for any LLM generated changes that tooling has a serious cost to it that cost is unrealistic for your hobby projects so your software is worse-by-default - **AdamRLucek** (2026-07-05): https://x.com/AdamRLucek/status/2073793164234617025 - No one wants to talk about how setting up agentic „loops” has almost nothing to do with the code being generated- rather effective tests, CI, scans, codebase rules, formatting/linters, documentation, historical references, QA tools, hooks, templates, tickets, reviews etc - **acoyfellow** (2026-07-05): https://x.com/acoyfellow/status/2073709485512593662 - visual-diff a framework-agnostic 3-tier cascade for deciding if two rendered UIs are the same: - DOM structure, - computed style, - and pixel diff all have to agree https://t.co/uk0UHPVauI - **RhysSullivan** (2026-07-05): https://x.com/RhysSullivan/status/2073664911893532753 - i think where i'm landing is that the code doesn't matter, but the types do - **thdxr** (2026-07-05): https://x.com/thdxr/status/2073621547718504628 - i feel like every debate on here breaks down into employed vs unemployed - **elithrar** (2026-07-05): https://x.com/elithrar/status/2073563196816834991 - July 4th, Burlington, VT. https://t.co/lcccPFUmR6 - **RhysSullivan** (2026-07-04): https://x.com/RhysSullivan/status/2073538819132207489 - who needs an MCP when you can just let your agent call the API get your agent connected to google search console with https://t.co/YqTmwo2dnx https://t.co/WkFxpYJOQN https://t.co/m9DFEcfeUK - **thsottiaux** (2026-07-04): https://x.com/thsottiaux/status/2073506936725581911 - @mark_k @OpenAI It's pretty simple. We are bringing ChatGPT to Codex so that Codex can be in ChatGPT in Codex in ChatGPT. - **southpolesteve** (2026-07-04): https://x.com/southpolesteve/status/2073488750823260460 - vinext started as an AI driven experiment: could one person + agents reimplement enough of Next.js on Vite to be useful? The answer was yes. But since that first release, the story has been the humans turning it into something real. https://t.co/F2tqdl3hyb - **mitsuhiko** (2026-07-04): https://x.com/mitsuhiko/status/2073488508816151038 - I had some vibes that Opus 4.8 was performing worse than older ones for some of uses that are off distribution and now I have the receipts. Latest Opus/Sonnet are causing tool invocation failures on Pi's edit tool when older ones did not! I wrote about it. https://t.co/96Qd8zfdVc - **matthewcp** (2026-07-04): https://x.com/matthewcp/status/2073473448718770531 - Water is the GOAT molecule. Best to ever do it. - **James_Elicx** (2026-07-04): https://x.com/James_Elicx/status/2073457440461303836 - Vinext 1.0 (Beta) is here! Next.js on Vite just got even better. Faster builds, smaller bundles, increased parity with Next.js, and improved cross-platform support. Transform any existing Next.js app to run on Vite with `npx vinext init` https://t.co/la1edKvtl0 - **ibocodes** (2026-07-04): https://x.com/ibocodes/status/2073397077598687303 - stop overcomplicating your stack, just deploy everything to cloudflare - workers (compute) - d1 (sql db) + better auth - kv (cache) - r2 (object storage) - queues (background jobs) - hyperdrive (db acceleration) - email routing (sending) - + waf & ddos built-in worst case your top bill is $10 - **thdxr** (2026-07-04): https://x.com/thdxr/status/2073238046296924466 - lately after a big change instead of reading the diff i ask the agent for a summary of what it did in each file anything weird sticks out immediately and 1-2 prompts later it's completely as i want it files + functions signatures i need to know, care less about function body https://t.co/9xsx4Dyblh - **odysseus0z** (2026-07-04): https://x.com/odysseus0z/status/2073205025204314138 - All the best agent frameworks are written by previous web framework authors. @flueai is written by @FredKSchott who did @astrodotbuild Eve by @vercel and @rauchg's team who did @nextjs @mastra by @calcsam who did @GatsbyJS. Agent framework is the new web framework - **Cameron_C2** (2026-07-03): https://x.com/Cameron_C2/status/2073143848977531024 - X at the speed of Rust, powered by Tanstack, Vite and Rolldown! https://t.co/PRQEGCxKPx - **KentonVarda** (2026-07-03): https://x.com/KentonVarda/status/2073125444048842914 - Wow, GLM-5.2 at home at 80tps for "only" $40,000. https://t.co/bFOCkP2Aaj Obviously still out-of-reach for most, but getting close to being cost-effective for the tokenmaxxers. A year ago there would have been another digit on this number. Where will we be next year? - **grinich** (2026-07-03): https://x.com/grinich/status/2073118730608152872 - WorkOS Launch Week - Day 5 🧩 Widgets API 🧩 Build any UI your product needs. WorkOS handles everything underneath. 🌐 Direct data access via browser-side APIs 🤖 Give agents a schema that doubles as the documentation ✏️ Customize with a full audit trail of user actions https://t.co/N9wBqjlQEI - **gokulr** (2026-07-03): https://x.com/gokulr/status/2073079633608900934 - Three non obvious things: 1. Find someone who can easily switch between strategy and execution, between the clouds and the dirt, between the board room and the daily standup, between the 60,000 foot view of the business and the lowest level of execution detail. I feel this is a non negotiable that people overlook. You want a doer, not just a talker, someone who hasn’t been lulled into the complacency of becoming a capital e “Executive” but can’t actually drive projects themselves. Whatever you do, don’t compromise on this. Execs are great talkers, so don’t get seduced by this. Drill into multiple levels of detail on their work and see if they understand it and can explain their work at that level. 2. Find someone who can deeply engage with both the product development team and the business / sales team. Yes, it’s likely they will lead one or the other, but they can’t be a single dimensional person who doesn’t understand product or doesn’t have a strong opinion on product. In other words, don’t hire a person who can’t truly engage with product and Eng as an equal. Have them meet 1:1 with key product leaders. Bring them to a couple of product sessions and see how they engage. Have a product centric dialogue with them. If you don’t get inspired or energized with their product vision or thinking, likely pass. 3. Remember that this is a bespoke role. The right near peer for each company / founder is unique and one of a kind. Eric, Sheryl, Keith, Christopher and Emilie are all very different from each other, both in styles of operation and personalities, but all wildly effective in their own way. Make sure your near peer is a fit for YOU and YOUR company. Understand their style of working by speaking with as many coworkers as possible. Make sure they pass the airport test with you. Meet with 5 founders who’ve successfully hired near-peers (and the near-peers themselves) to undestand the hiring process and how they figured out this person was “the one”. - **elithrar** (2026-07-03): https://x.com/elithrar/status/2073060157785821647 - in some personal news: after 10+ years here in the US, i’m now a permanent resident 🦅 🇺🇸 what an incredible wave of relief to have read “case approved”. - **zeeg** (2026-07-03): https://x.com/zeeg/status/2073059050447941969 - every thing ive tried turns 1-2 hour HITL session into a 6+ hour session where I have to go back and do a massive amount of inspection its a better model, but idk what loops its closing. the cost and latency aren't acceptable. my first session was $400. so maybe its closing anthropic's margins? maybe its better if you avoid entirely their attempted coordinator pattern and just use it for specs/verification, but it doesnt seem to be making complex changes any easier for me - **GeoffreyHuntley** (2026-07-03): https://x.com/GeoffreyHuntley/status/2073040942282469785 - me: you can prompt llms to do mutexs btw. also me: shuddering in thought that aiwf next year is gonna be about mutex engineering - **gokulr** (2026-07-03): https://x.com/gokulr/status/2072856480336986393 - GROWTH FOUNDERS: HIRE A NEAR-PEER A portfolio founder approaching $100M in ARR asked me about the single biggest and most impactful hire they can make. I thought for a bit tand said: "Hire a near-peer". In every generational company I've been part of, the founders hired a near-peer, who was essential to the company's success. - Google: Larry and Sergey hired @ericschmidt . - Facebook: Mark hired @sherylsandberg . - Square: Jack hired @rabois . - DoorDash: Tony hired @chrispa - Coinbase: Brian hired @emiliemc Characteristics of a near-peer: 1. They're so good that the founder will be fine reporting to them if the roles were reversed. (Mark has said publicly that he'd be fine reporting to Sheryl) 2. Their strengths perfectly complement the founder's strengths; however, they share many cultural attributes with the founder and pass the founder's airport test, since the founder will be spending a ton of time with them (example: Eric being a Computer Scientist, which was culturally very important at Google back in the day) 3. They're systems builders who have already operated at the scale you're growing into. Pattern recognition on $100M to $1B is not something you build in real-time. A near-peer lets the founder focus on what only the founder can do (product, vision, culture). The near-peer handles everything else. Also, near-peers stay for the long haul. Decade-plus tenure is standard. The reason this hire matters more than any other: after $50M ARR, the bottleneck shifts from product-market fit to organizational scale. The founder is still the visionary. But the company needs someone who has already scaled a company of that size. Most founders wait too long. They hire functional VPs first (Sales, Marketing, Engineering, Finance) and hope the collective covers the gap. It rarely does. A stack of VPs reporting into a founder who has never scaled a company creates coordination overhead, and the founder becomes the bottleneck. The near-peer absorbs that overhead. They turn the founder's vision into daily operating decisions. They give the VPs a leader who has actually run a company this size. Growth founders: if you're between $50M and $300M in ARR and every week feels like a coordination tax, the near-peer is the hire that determines whether you build a $10-100B company or top out at $1B. Start the search now, ideally through warm intros from your venture/angel investors and advisors. - **zeeg** (2026-07-02): https://x.com/zeeg/status/2072764730012516667 - Claude Tag is a terrible product name. Sorry it needed to be said. - **Cameron_C2** (2026-07-02): https://x.com/Cameron_C2/status/2072725709764677918 - oxlint keeps getting faster! @yagiznizipli has been running oxlint on the X codebase to find hotspots and make it faster! https://t.co/HeBlqerzDP - **grinich** (2026-07-02): https://x.com/grinich/status/2072712604678754383 - WorkOS Launch Week - Day 4 🔒 Step-up Auth 🔒 Also known as "sudo mode" ⏱️ Check the freshness of a user session 🚨 Gate sensitive actions on recency: account deletion, billing, API key management ↩️ Re-auth through AuthKit, return users exactly where they left off Read more: https://t.co/UDSw7eKeSB - **pheoru** (2026-07-02): https://x.com/pheoru/status/2072707761331212617 - Cloudflare Mesh now supports hostname-based routing for private traffic and as exit nodes. Reach an app (e.g. wiki.internal.local) even when its IP is unknown / ephemeral. And no, you don't need to host DNS servers, changing hosts file will do. https://t.co/jC3md6XKX0 https://t.co/7hRXs6FHFQ - **boshen_c** (2026-07-02): https://x.com/boshen_c/status/2072574230378815545 - ??? https://t.co/JBqwv1LsIJ - **steipete** (2026-07-02): https://x.com/steipete/status/2072532278476148881 - Never thought I give @Steve_Yegge a shoutout. He was just early, like most visionaries. Now everyone is building factories. - **mattsimpsn** (2026-07-01): https://x.com/mattsimpsn/status/2072380106383761421 - durable objects memory metrics are live. track usage per object across percentiles and catch leaks before they become errors workers metrics got deployment markers too. no more guessing which release caused a regression what’s next on your wishlist? dms open https://t.co/YHLlLoX7YF - **BraceSproul** (2026-07-01): https://x.com/BraceSproul/status/2072375136368660515 - https://t.co/1Gjh2SUHt8 - **frolic** (2026-07-01): https://x.com/frolic/status/2072355384828821898 - I have always wanted something like this, to be able to build small APIs/services but let something else handle the auth and/or payments. https://t.co/Fu27CDNwz2 - **rough__sea** (2026-07-01): https://x.com/rough__sea/status/2072351392287825952 - It has taken longer than expected, but wasm is finally becoming important - **IkeShehadeh** (2026-07-01): https://x.com/IkeShehadeh/status/2072349675429273890 - Yo, @KingJames. Sign with the @warriors & not only will we design a sandwich together and eat FREE @ikessandwiches for life, but I'll put your face next to mine just down the street from Chase center. Let's wreck this league. https://t.co/hMe4r1TH6A - **hyfdev** (2026-07-01): https://x.com/hyfdev/status/2072326330738504155 - Do you know the TypeScript compiler bundled by Rolldown runs ~3x faster than Webpack's? https://t.co/Jw26G6vd6F - **mattzcarey** (2026-07-01): https://x.com/mattzcarey/status/2072219893059965323 - MCP typescript SDK v2 is now in beta!! try it today: npx @modelcontextprotocol/sdk@latest codemod upgrade-v2 - **QuinnyPig** (2026-07-01): https://x.com/QuinnyPig/status/2072132613196898714 - Okay I owe my @OpenAI friends an apology for sleeping on Codex. I was not aware how strong your game was. This is... really quite something. - **SukritGanesh** (2026-06-29): https://x.com/SukritGanesh/status/2071716654883737884 - California should legit take control of the city of San Francisco, suspend every single local regulation (not just for housing, but for everything) for 10 years, and hand control back once the mess is sorted out. https://t.co/d4Q31XgKuK - **boristane** (2026-06-29): https://x.com/boristane/status/2071707504996946004 - get agents that look at everything in your app in real-time and fix all issues as they happen https://t.co/AaSkcOOYeD - **DKThomp** (2026-06-29): https://x.com/DKThomp/status/2071649087649128461 - Smart post. A chilling thought I've had recently, watching Mag7 cash flow crashing toward zero and watching AI bubble fears reemege, is that the US built its tech dominance on industries where the marginal cost of the product was basically zilch, and this happened in the context of a US financial media that loves panicking and talking about bubbles and doom at the earliest sight of an earnings miss. But with AI, you've got a capex arms race that seems like itll reward patient and loss-tolerant capital, and the business cycle might go thru some ups and downs that could trigger more AI bubble panics and encourage big US companies to pull back spending. Meanwhile what does China love? Ooutspending and out-losing its global private-sector industrial competitors for years and years ... and the CCP doesn't change its investment strategy based on compelling CNBC clips. I think the US is winning on AI right now, and I'd like that to remain the case, but with rising inflation, data center populism, and other resource limitations, I don't think it's crazy to think that the US has some real challenges. - **BraydenWilmoth** (2026-06-29): https://x.com/BraydenWilmoth/status/2071620304602857894 - Might be a little slow to reply for a bit 😅 https://t.co/Qvf7PVZ7t7 - **yrechtman** (2026-06-29): https://x.com/yrechtman/status/2071598731452129597 - Obviously yes. https://t.co/eHSbNDf7qx https://t.co/eXHVdZW39W - **grinich** (2026-06-29): https://x.com/grinich/status/2071433948870430890 - biked to inverness and back ☀️🚴‍♂️ turns out you don’t need a car for a weekend trip! https://t.co/RWrB89cAlT - **ankrgyl** (2026-06-28): https://x.com/ankrgyl/status/2071329164125147440 - there are two interesting mega trends right now * rework my company to operate with agents * rework my product to be an agent requires a ton of discipline not to conflate these. not much room for horizontal agent platforms - **sarukun99** (2026-06-28): https://x.com/sarukun99/status/2071271285007724650 - flue良さそうなのと。 仮想ファイルシステムをDurableObject上につくるのすごいな。 https://t.co/ZneHgZEg0m - **hyfdev** (2026-06-28): https://x.com/hyfdev/status/2071111504569712999 - Been digging into perf on https://t.co/ixnRFx5jSv lately, and the early numbers are wild. - Production build time drops 32–52% - dev-server cold start 67–83% and it scales! the bigger the app, the bigger the win. https://t.co/J8OAdgp6ix - **mattzcarey** (2026-06-28): https://x.com/mattzcarey/status/2071079007915819231 - @RhysSullivan This https://t.co/nUVvRkaxNN - **dillon_mulroy** (2026-06-27): https://x.com/dillon_mulroy/status/2070983687785914822 - i don't want to use your products agent, i want my agent(s) to be efficiently enabled to use your product https://t.co/APiSJHHTwX - **_dylanga** (2026-06-27): https://x.com/_dylanga/status/2070841285468966999 - Inspect has a set of tools that, if they error, will return a “hint” as part of the tool. The next turn will use the hint to help steer it to the right usage. https://t.co/CcwU0qAGRt - **southpolesteve** (2026-06-27): https://x.com/southpolesteve/status/2070676424122356172 - I really want CLI tools to start speaking directly to coding agents with errors. Instead of: “Error:” Return: “Here’s what broke. Here’s how to investigate. Here’s how to generate a redacted repro. Here’s where to send it.” Obviously there are privacy/security challenges to solve. But if we get this right, we can have a virtuous software improvement loop. Every failed agent interaction becomes a high quality bug report. Agents find the bugs. Agents fix the bugs. Software gets better faster. - **corywilkerson** (2026-06-26): https://x.com/corywilkerson/status/2070624948801183983 - models. vibes. edge compute. tiny demos. huge network. excellent teammates. interlinked. apply. https://t.co/2TfpIgdXDR - **elithrar** (2026-06-26): https://x.com/elithrar/status/2070583599238164627 - FYI: ArtifactFS now supports fully async clones (incl. refs!), meaning the cost to opportunistically mount multiple repos into a sandbox is near-zero. by default and without the --async flag, refs are sync (so the file structure is there), and blobs are async. https://t.co/nrVhdqg6OF https://t.co/WhtvJDCR1u - **justinstorre** (2026-06-26): https://x.com/justinstorre/status/2070576734173643002 - Anthropic - “AI should be government regulated” https://t.co/MhQhVUKbCX https://t.co/sPZbTVWKhr - **OpenAI** (2026-06-26): https://x.com/OpenAI/status/2070555272230384038 - Introducing a limited preview of GPT-5.6 Sol, our next generation frontier model, as well as GPT-5.6 Terra, a balanced model for efficient, everyday work, and GPT-5.6 Luna, a fast and affordable model for high-volume work. https://t.co/OoM83SyISN - **jshchnz** (2026-06-26): https://x.com/jshchnz/status/2070533022223667672 - When your CFO looks at the Datadog bill https://t.co/YWJvIUd5zA - **mattzcarey** (2026-06-26): https://x.com/mattzcarey/status/2070531713571143702 - @seflless @RhysSullivan My infra is fine to be orphaned. Some teams make billions of Durable Objects per day. No issues - **thdxr** (2026-06-26): https://x.com/thdxr/status/2070508326623047872 - mcp was clearly inspired by looking at lsp and all the progress in the spec has been slowly realizing it should be nothing like lsp - **dillon_mulroy** (2026-06-26): https://x.com/dillon_mulroy/status/2070479288420258279 - fokosdb is probably the thing i’m most excited about in cloudflare ecosystem atm what is it? what if dynamodb, but cloudflare https://t.co/KDe1lIum78 - **verrsane** (2026-06-26): https://x.com/verrsane/status/2070438648638587287 - @RhysSullivan Would u be interested in how glean solved this? It’s orthogonal to code mode - **mattzcarey** (2026-06-26): https://x.com/mattzcarey/status/2070438382589657533 - I shipped some docs for stateful codemode :) https://t.co/ubFLpqqEB7 - **bgurley** (2026-06-26): https://x.com/bgurley/status/2070427385237741797 - This is what’s causing Anthropic to aggressively beg for govt protection (see below). Customers are finding cheaper alternatives. Keeping employees requires continuing ultra-rich secondaries ($$$) that are dependent on revenue growth. When you can’t win on the field go to DC. https://t.co/fbO2MfiWHu - **mattzcarey** (2026-06-26): https://x.com/mattzcarey/status/2070413485196898627 - @RhysSullivan Stateful codemode - **bgurley** (2026-06-26): https://x.com/bgurley/status/2070405731396776168 - Yes. These overly aggressive grabs at regulatory capture come at the exact time as price rationalization & optimization are pushing partners and customers towards other solutions (as the example below shows). Hope the govt knows they are being manipulated. https://t.co/fq1UFO1q06 - **RhysSullivan** (2026-06-26): https://x.com/RhysSullivan/status/2070401257693712536 - tbh i wonder if we're overindexed on codemode the way we were with standard tool calling & then bash after that code mode is the best approach so far, i think you can call anything from it and you get return types, chaining, filtering, but wonder if there's something more - **RhysSullivan** (2026-06-26): https://x.com/RhysSullivan/status/2070311929038680262 - https://t.co/4dZadvkZoi - **avenceslau** (2026-06-26): https://x.com/avenceslau/status/2070306371539464443 - With rollbacks you can do one of my favorite things, HMD (health mediated deploys). And when something fails you just rollback It could be this easy, I have been having too much fun with specialized workflows https://t.co/rzP8dzk5ri https://t.co/xed0fsqOD7 - **jxnlco** (2026-06-25): https://x.com/jxnlco/status/2070237290346222034 - @MichaelArnaldi @jlongster I think I saw the slack I’ll close the loop after some work stuff. - **simonw** (2026-06-25): https://x.com/simonw/status/2070212662903099536 - Also not a great advertisement for a sandboxing product: @daytonaio effectively saying they don't trust the security of their product enough to expose the source code https://t.co/B6cinfTl01 - **eglyman** (2026-06-25): https://x.com/eglyman/status/2070137496399155243 - Today @karimatiyeh and I are both taking new titles as Co-CEOs of @tryramp. If you know us, this won't feel like a change. From when we first started building together twelve years ago, our partnership has run on a couple of motivating principles. On decision-making, we trust each other completely to make critical calls for the company across every function. And on organization design, technology is not a distinct part of the company - it is the entirety of it. That is why Karim has for years directly managed risk, operations, and marketing. Most importantly, at Ramp there is no line between the people who build and the people who do everything else. Everyone is a builder. For the last 2,656 days, we have run the company this way. This only makes it formal. We thought it was important to do it now because of how we see the AI exponential reshaping what Ramp can be. Decisions of company strategy are increasingly decisions of technology and systems design. We have always believed every function should be approached as a systems-engineering problem (even when the system was primarily human) but the rise of machine intelligence makes this existential. Every part of the company must be positioned to leverage the continued explosion in model intelligence and capabilities. If we do this well, each step-change in what models can do compounds automatically into better products and faster execution without anyone having to rebuild the company to capture it. If we fail to operate this way we will ultimately be outcompeted by a new company that does. We are also making Rahul Sengottuvelu our CTO. @rahulgs has led Applied AI at Ramp since joining us three years ago through the acquisition of his prior company, Cohere. Before that, his first company was building customer-service agents on GPT-3 at a time when almost no one knew what a large language model was, and he has spent every year since pushing the frontier of what existing models can do. He has also been right on nearly every major technical direction in AI well before it was obvious. Building Ramp now means applying AI to every part of it, and Rahul is the person stepping up to lead that work. We are still very early in the history of Ramp. Our current chapter is perhaps the most dynamic, but we have never been more optimistic on where it is going and the mission has never been more important. The businesses that trust us are navigating the same shift we are, and we intend to be there for all of it: managing their token spend, supercharging their finance teams, and helping them get more out of every dollar and hour. - Eric & Karim - **avenceslau** (2026-06-25): https://x.com/avenceslau/status/2070133934269055305 - No more rollbacks wen, amazing job @wishee0 https://t.co/OFx988bRaN - **ankrgyl** (2026-06-25): https://x.com/ankrgyl/status/2070004157034188919 - modern evals are about optimizing tokens, turns, cache hits. what does it take to achieve X, not can this model achieve X? - **swyx** (2026-06-25): https://x.com/swyx/status/2069964772003770673 - lots of folks prepping talks next week (congrats!). Some thoughts from RLing on thousands of hours of engineer- and researcher- focused talks: - AI generated svgs > AI generated imgs. MAXIMUM 4 ai slop images in your slides, I don't care how pretty your mom thinks they are (exception ofc if your talk is ABOUT imagegen) - Be pointy. Better to have 1 message with 5 surprising applications, than 5 messages with no concrete examples. - Put code on screen. Engineers like to see code. Especially if they can nitpick it to death over things that aren't the point. - Don't forget to Entertain. Being actually funny, or good at telling relevant anecdotes, is more important than adding yet another bullet point. - Have a Thesis. every talk gets ONE "if you remember one thing from this talk, it is this" card. 1) SO MANY PEOPLE DON'T USE IT. 2) SO MANY PEOPLE DON'T PLAN FOR IT. you get one. use it well? - Have a Thesis Slide. you've been in that talk where everyone gets their phone out to take a photo of the slide. Because people see 1000x more images than videos, you are much more likely to have a viral talk/slide if you have a single viral slide. You are much more likely to have a viral slide if you TRY and most people do not TRY. Struggling? Collect examples from talks you like and adapt their format to your thing. Your slides have a power law - don't spend 5% of your time each on 20 slides, spend 80% of your time on 1 slide and have the rest build up to that 1 slide. - You might not need slides. live demo in IDE, off the cuff rambles, pull audience member up to roleplay, do call and response with the audience, sing/perform, voice over vibe videos, I have seen it all. higher risk/effort, higher reward when done well. - Be pleasant to listen to. Talks with bad/no visuals still can be listened to. Talks with bad audio are DOA. Be confident, project, remember vocal variation, evoke emotion. - Design the emotional journey. Start strong, end strong, have a peak aha/laugh/thesis moment in the middle. Everything else is buildup. - Data driven talks are underrated. Present pretty charts and surprising, authoritative numbers. Use the stage authority to infer from the data to support the broader thesis. Done well, it will feel like my obvious conclusion from objectively looking at the data you presented, not you feeding the conclusion to me. - How to shill your product/company without feeling salesy: teach me everything I didn't know I should to know about the problems you solve, and then you shall have earned the right to convince me you are the guys to trust to solve it once and for all. - Actively watch a lot of talks. like with anything you both need a lot of reps AND you need to explore to find the styles/role models that you shine in. Passive = no introspection after watching, just mindlessly autoplay next video. Active = trying to articulate why a talk was good/bad after watching. If you want to REALLY hone it, think about how you feel about a famous talk and then TRANSCRIBE the talks and read the words on the printed page, and then compare with a "normal" talk and define rules you will follow for yourself to improve. thanks to @dexhorthy for organizing the AIEWF Speaker prep meetup tonight. we should actually do more of these.... - **FredKSchott** (2026-06-25): https://x.com/FredKSchott/status/2069961771650277759 - @skeptrune @astrodotbuild Can’t imagine a better use case for this than docs ⚡️ - **astrodotbuild** (2026-06-25): https://x.com/astrodotbuild/status/2069960027025093018 - Introducing Sätteri — High-performance Markdown & MDX processing for the JavaScript ecosystem. https://t.co/nLkscI6l6H https://t.co/urZKMpCQHe - **samlambert** (2026-06-24): https://x.com/samlambert/status/2069923372994306216 - To survive in a fast moving asynchronous work culture you need to be able ingest written information incredibly quickly. You also have to be good at ignoring things. - **dillon_mulroy** (2026-06-24): https://x.com/dillon_mulroy/status/2069859152881406132 - free idea if anyone wants to build it: 1. ast-grep outline 2. visualize w/ dashboard like this 3. ??? 4. profit https://t.co/hruPuAlMAu - **PatrickToulme** (2026-06-24): https://x.com/PatrickToulme/status/2069857412660801738 - A few thoughts on OpenAI's Jalapeño chip announcement today: 1. This chip is most likely the first one virtually entirely developed by Codex/GPT. Codex with whatever internal coding model (GPT 5.6/6.0 whatever) coded the entire software stack and most likely the hardware design 2. OpenAI will write all of their inference serving in pure Jalapeño ISA (instruction set architecture). Why? They only need to get say a few production models serving on Jalapeño. They can handwrite with Codex the entire model in pure ISA to get very high performance 3. They are most likely running Codex/GPT in custom RL envs to teach the models direct Jalapeño chip programming at ISA level 4. This is a massive cost savings for OpenAI and only possible IMO due to the breakthroughs in agentic coding. An AI company with frontier coding models can now become a hardware vendor with only a small team of experienced SWEs and an infinite amount of tokens This is the first chip program fully accelerated by frontier AI. - **saammotamedi** (2026-06-24): https://x.com/saammotamedi/status/2069829251650732509 - Congratulations to @clattner_llvm, @iamtimdavis, and the entire @Modular team on today's announcement that Qualcomm has agreed to acquire the company for ~$4 billion. We at @GreylockVC have been fortunate to back Chris, Tim, and the team since their seed round, and it's been a privilege to partner with them from the start. Chris and Tim are an incredibly rare combination of world-class systems thinking and product taste, that can take deep technical complexity and make it highly usable for millions of developers. When they started Modular a little over four years ago, they had a prescient thesis that AI needed an open platform to runs models at state-of-the-art performance across heterogeneous hardware, without requiring developers to rewrite anything. They delivered it with Mojo and MAX, and now they'll accelerate that mission by pairing their software with Qualcomm's silicon leadership. We're grateful to have invested alongside @davemuni @GVteam, @deepnishar @generalcatalyst, and Chris Ré, and we're excited for this important milestone for Modular. - **irl_danB** (2026-06-24): https://x.com/irl_danB/status/2069789821996781732 - > the legend tweets something 6-24 months after it is conventional wisdom and well-trod in certain circles > the world starts paying attention I can count at least 4 times this has happened with Andrej specifically, likely more. distribution really is everything! your ego thinks this is a blackpill: why should they get credit for the thing others have been doing for years? but it's actually a whitepill: your alpha has a 6-24 month lead time, anon - **boshen_c** (2026-06-23): https://x.com/boshen_c/status/2069449703935336846 - We withdrew the Rust React Compiler integration from Rolldown and Vite because it increased the binary size from 28.7MB to 33.8MB, a 17% increase, which cannot be justified for all Vite users. My current experiment is to have Claude understand our constraints. Preliminary results show up to a 2x performance improvement with only a 1.4MB increase in binary size. Next step is to de-slopify and get the code into a maintainable state, and also upstream some of the changes. - **mattzcarey** (2026-06-23): https://x.com/mattzcarey/status/2069441064100843672 - protocols are great https://t.co/taVJN737rO - **boshen_c** (2026-06-23): https://x.com/boshen_c/status/2069412529982627910 - > Experimental Bundled Dev Mode It's finally here! We’ll make sure it becomes fully stable as soon as possible. https://t.co/upN84LXBRd - **unsafemut** (2026-06-23): https://x.com/unsafemut/status/2069396079905509884 - Give it a try 🙌 https://t.co/GFbzuV02ej - **_h_ana___** (2026-06-23): https://x.com/_h_ana___/status/2069395618154557911 - months of work. full bundle mode, it’s finally here, just as i promised https://t.co/FGdcg7HOGk - **vite_js** (2026-06-23): https://x.com/vite_js/status/2069389513135096218 - 🚀 Vite 8.1 is out ⚡ Experimental Full Bundle Mode: ~15x faster dev server startup & ~10x faster full reloads on large apps 🧩 Chunk Import Map to keep chunks stable & improve caching 🦀 WASM ESM integration 💡 Lightning CSS on the road to default 👀 and more! Blog post below. https://t.co/73yi9azjo9 - **samlambert** (2026-06-23): https://x.com/samlambert/status/2069257426759725208 - people are too overconfident in the belief that slop will overwhelm companies using too much AI. smarter models can refactor too. - **naval** (2026-06-22): https://x.com/naval/status/2069200779408474361 - Good marketing is finding different ways to tell the truth. - **elithrar** (2026-06-22): https://x.com/elithrar/status/2069199926639116667 - migrated workflows to Flue's "Durable Agents" API, which can use Node (scoped to process lifetime) or Cloudflare Workflows (persisted, recoverable) under the hood. next up: implementing a backend that doesn't rely on GitHub Actions w/ the Sandbox API. https://t.co/mmnEq48jGE https://t.co/mzdlrb01Wq - **matthewcp** (2026-06-22): https://x.com/matthewcp/status/2069178233652723890 - Core team member Princesseuh built Sätteri, a modern Markdown/MDX processor, inspired by what @lightningcss did for CSS. In addition improving perf, Satteri also has the most common remark/rehype plugins built-in. It's now the default processor in Astro 7. - **matthewcp** (2026-06-22): https://x.com/matthewcp/status/2069178229827502356 - For Astro 7 we wanted to tackle the problems of scaling larger static site builds. We decided to do like @vite_js and port as much to Rust as possible. https://t.co/RdimOu7j8g - **zeeg** (2026-06-22): https://x.com/zeeg/status/2069176778049138798 - i appreciate a company that attempted to exfiltrate secrets from our customers machines is running a PR campaign as if they're not running a blackhat operation https://t.co/9D2l27hWzF @janakiramm @thenewstack word of advice, unpublish things you're claiming as facts that you did not verify (and yes, prompt injections are a real risk with agents, news at 11) - **superhighfives** (2026-06-22): https://x.com/superhighfives/status/2069173970222686442 - Come work with me and @minglu and @corywilkerson! It’ll be great, I promise. https://t.co/hqGxdlvSQe - **ritakozlov** (2026-06-22): https://x.com/ritakozlov/status/2069172171650318806 - @kentcdodds @Cloudflare @threepointone we have supported websockets for years now? - **corywilkerson** (2026-06-22): https://x.com/corywilkerson/status/2069169732922872149 - Hey folks! We’re hiring an excellent hands-on technical manager to lead AI Gateway here @Cloudflare. The job description says it all: deeply technical, deeply hands-on, and a love for dev tools and the people building them. You’ll be in the code, in the customer conversations, in the harnesses and agents that build the product alongside the team. I’m a little biased but this thing is taking off rapidly and we’re doing it in a way only Cloudflare can. We’ve got all the right pieces: the culture, the team, the product…we’re just missing you! https://t.co/PYmZpBMfq3 - **ankrgyl** (2026-06-22): https://x.com/ankrgyl/status/2069162526806573200 - we've put a ton of work into modernizing the waterfall view in braintrust. i'm most excited about how obvious we make cache behavior (blue is cached and purple is uncached). in the era of vibe coded ui, this one actually took some serious human brainpower to get right. https://t.co/Sa4s1oqETe - **badlogicgames** (2026-06-22): https://x.com/badlogicgames/status/2069156845940244647 - "human review at the center" they are getting it. https://t.co/nWK28DmJ2L - **matthewcp** (2026-06-22): https://x.com/matthewcp/status/2069147909547782570 - Any Astro app can be controlled by Hono new. Create `src/fetch.ts` and handle the full request pipeline using Hono's ecosystem. https://t.co/RdimOu7j8g https://t.co/fb7jPPvOHA - **southpolesteve** (2026-06-22): https://x.com/southpolesteve/status/2069144590536343667 - The new Hydrogen preview works out of the box on vinext too https://t.co/3CNOMQRLUh Excited to see where this goes! https://t.co/K4p2xJcTKU - **ivictorialiao** (2026-06-22): https://x.com/ivictorialiao/status/2069136578963026223 - CHRISTIAN AT PLANETSCALE DOT COM https://t.co/ANXw2aBpHV - **_heyglassy** (2026-06-22): https://x.com/_heyglassy/status/2069133492840194340 - Today is my first day @PlanetScale as a Developer Educator. It hasn’t quite sunk in that I work at the database hat company I’ve been a fan of for years. PlanetScale is the gold standard for infrastructure companies; there have been many days when I’ve worn the database hat and tried to emulate their spirit in my work. I’m excited to join this set of heretical, truly p99 folks. You could spend an hour bumping into people at the office and learn that the keyboard rack was handmade by an engineer who flew in from the East Coast to install it, or that your office manager is a sake sommelier with a really granular rating spreadsheet, and that wouldn’t even scratch the surface of how great every single person is. I always have a great time nerding out about databases, and I’m excited to bring more of that joy to you through events and content, or over coffee. If you want to chat, please say hi in my DMs or at christian at planetscale dot com. - **mwylde** (2026-06-22): https://x.com/mwylde/status/2069124825730490786 - \1 We recently announced GA pricing for the @CloudflareDev Data Platform, our R2-backed real-time data lake, which has been free during beta. But beyond the actual numbers, I want to talk a bit about our pricing philosophy - **PlanetScale** (2026-06-22): https://x.com/PlanetScale/status/2069108764167549192 - Learn more about using Hyperdrive with PlanetScale https://t.co/0jONdEsn5h - **OpenAI** (2026-06-22): https://x.com/OpenAI/status/2069104288417390688 - Patch the Planet is our effort to help open source maintainers move from security findings to merged fixes. We’re working with Trail of Bits, HackerOne, Calif, researchers, and maintainers to bring Codex Security and advanced models into the remediation process, with human review at the center. - **FredKSchott** (2026-06-22): https://x.com/FredKSchott/status/2069102152606523502 - A new Rust [...] a new Rust [...] and more [Rust.] 🦀 https://t.co/ExpYc1ouXv - **astrodotbuild** (2026-06-22): https://x.com/astrodotbuild/status/2069098646604304464 - Astro 7 is here! A new Rust compiler, a new Rust Markdown/MDX processor, Vite 8 and more. Get ready for 60%+ faster builds. https://t.co/RG0CPr6IUF - **flueai** (2026-06-22): https://x.com/flueai/status/2069087745054790008 - $ flue docs Fully local, offline, versioned docs. Fully searchable. Designed for agents. https://t.co/ox56TkrS2R - **badlogicgames** (2026-06-22): https://x.com/badlogicgames/status/2069051109558083924 - People of https://t.co/TgG5bkXUdV. The next releasrle if pi-ai (not the coding agent) will be a breaking release. To be support consumers like @FredKSchott wonderful flue, you can now selectively pull in only the SDK impls you need. This can reduce bundle size. It also removes global state. And it allows you to plug in your own env resolution and credential storage. The old API is still available via the @earendil-works/pi-ai/compat module. See the updated README.md of the pi-ai package for details. https://t.co/bqwLSTjV0j The CHANGELOG will be updated with a migration guide once the remaining work (coding agent and new agent harness integration) is complete sometime later this week. - **thdxr** (2026-06-22): https://x.com/thdxr/status/2069042956426760492 - OpenCode Console is going to look good now https://t.co/Y79wg1Gv1D - **LakeAustinBlvd** (2026-06-22): https://x.com/LakeAustinBlvd/status/2069032778612482166 - So many interns starting today in @Cloudflare Lisbon that we had to convert common spaces into desks. https://t.co/Bn4JWI0gK8 - **southpolesteve** (2026-06-22): https://x.com/southpolesteve/status/2068877355024515219 - Going to start a contrarian thesis fund. Only writing checks to people who go to Golden Goat https://t.co/HCDCD9lLpW - **lessin** (2026-06-21): https://x.com/lessin/status/2068831642278568026 - Kickstarter... i always feel burned by you... but fine let's try again because this looks great! https://t.co/ktu9BMYq1l - **fortyfivan** (2026-06-21): https://x.com/fortyfivan/status/2068823383232467445 - @briannekimmel Impressive! Did you get the whole atom Maia Racional backstory? - **dwr** (2026-06-21): https://x.com/dwr/status/2068812741943402889 - Feels like there’s significant latent demand for 70mm IMAX seating in major metros. Surprised there aren’t more being built despite the cost. Future is barbell: watch on your own device or watch in best possible theater format. - **briannekimmel** (2026-06-21): https://x.com/briannekimmel/status/2068786920612889044 - Brazilian friends, did I do good? A highlight of Rio was spending an hour in the record store listening to tracks on each album and giving the owner a thumbs up or thumbs down. I’ve had Di Melo, Joutro Mundo, Rosinha de Valença for years, a gift from my neighbor in Bondi Beach. https://t.co/ceNvAGMZGF - **eliakuratli** (2026-06-21): https://x.com/eliakuratli/status/2068727720243089498 - https://t.co/mzcl1Fg3Yw - **threepointone** (2026-06-21): https://x.com/threepointone/status/2068718370581536816 - a fully interactive "recovery lab" where you simulate different kinds of agent failures/recovery built with opencode + glm 5.2 (176k tokens, 3.50$), 1 shot after I fed it my 4k word slop analysis and the agents sdk repo pretty fkin cool. will polish and publish this soon. https://t.co/rrnjKm6zPu https://t.co/iyT9LEuGsC - **rough__sea** (2026-06-21): https://x.com/rough__sea/status/2068704065257910655 - what's interesting now is to create software that was not possible before - simply too expensive. implementing massive API surfaces, language ports, etc. this is the era of colossal-scale software - **TheAlexLichter** (2026-06-21): https://x.com/TheAlexLichter/status/2068667982373433600 - A little side quest appeared! With the proposed change in React itself, apps using Vite's RSC plugin would receive an LCP boost without any code changes in the actual apps. Reason is better resource scheduling via the added fetch priority. https://t.co/GV1q6t1OZj - **thdxr** (2026-06-21): https://x.com/thdxr/status/2068664567085818128 - a shortcut for this one is remember this rule: if you're fixing someone else's product you should probably work on something else - **thdxr** (2026-06-21): https://x.com/thdxr/status/2068664103917240757 - this is a good example to learn from since this product is standalone, the aha moment requires a user to connect 2-3 services before they see the value the flow for doing this is often difficult particularly with their most useful data that is unique to them very few people will make it through onboarding to get to this point and on top of this, this is basically an extension to someone else's product that the user already has the aha moment with it will be 1000x easier for them to incrementally get their users to do this thinking a bit will save you years of pain working on the wrong thing - **avenceslau** (2026-06-21): https://x.com/avenceslau/status/2068661232991699300 - I wonder how many people are missing something like this: - Define your CI as a workflow, you can run any TS command and can run your pnpm commands via step runner (or something like that) - You can build all of this at cloudflare in userland it's just not that obvious How does this work? - I have one worker on my account that uses Dynamic Workflows and passes in capabilities that expose the github API (or Artifacts if you use them as a remote), a wrapper around Cloudflare's API. - All of my repos/workers define a CIWorkflow, on push you either have a github app push the event to a queue (the queue is also subscribed to Artifacts events) - On an event the queue's message kicks of a workflow that will bundle the CIWorkflow on a worker and send it to the worker that runs the Dynamic Workflow. My repos only have to have ci.ts file that extends CIWorkflow every binding that I add to my CIWorker can be used in any of my other repos. Should I build this into Cloudflare so that no one has to this lengthy setup? - **ibocodes** (2026-06-21): https://x.com/ibocodes/status/2068652359194992778 - hate to break it to you but most of your projects don't need vps + postgres or supabase + vercel just throw everything on cloudflare workers + d1 $5/month until you hit 10k users stop overengineering https://t.co/CxwFpRLnXu - **cnakazawa** (2026-06-21): https://x.com/cnakazawa/status/2068629379723444584 - A wild Cloudflare employee appeared contributing to React. nbd https://t.co/c65wK5QqJO - **RhysSullivan** (2026-06-21): https://x.com/RhysSullivan/status/2068506354583851365 - niche complaint but why are google oauth scopes such a mess if you select the wrong combination their auth just fails completely, you can also accidentally select a narrowing scope which breaks your broad scopes anyone else dealt with this - **makisuo** (2026-06-20): https://x.com/makisuo/status/2068475175142924644 - Maple now has OTEL service.namespace support and will automatically group any services with other services in the same namespace. https://t.co/FJV4pm6uJE - **reach_vb** (2026-06-20): https://x.com/reach_vb/status/2068452487468564742 - @irvinebroque yes! @AriX and team cooked fire with that!! - **blader** (2026-06-20): https://x.com/blader/status/2068433727248847298 - codex is now my primary interface into both gmail and linear a harbinger of what is to come for established web interfaces - **dexhorthy** (2026-06-20): https://x.com/dexhorthy/status/2068364484306247911 - you should have a linter Hands down You should have detailed rules, you should push determinism as far as it can go Use ast analysis to tell your coding agents what needs to be fixed You should absolutely do this BUT If your anti-slop strategy is an LLM and a handful of linters You’re gonna be disappointed - **thdxr** (2026-06-20): https://x.com/thdxr/status/2068350101920067738 - i've been building things for developers for a long time which means i have a lot of experience with api design but LLMs are making me reset a bit instead of neat simple APIs it's probably worth trading that off for more powerful APIs - **flueai** (2026-06-20): https://x.com/flueai/status/2068344872549925347 - NEW: Schedules Run daily summaries, recurring reports, data synchronization, and more. Schedules let your Flue agents run automatically, even when no one is actively prompting them. https://t.co/5SYwA47gAR https://t.co/uLNt1DHGWU - **kevin_flansburg** (2026-06-20): https://x.com/kevin_flansburg/status/2068289785232564293 - If it truly doesn’t make sense to run a test, that should be mathematically provable given my specific data, and not decided based on vibes or at the population level. - **kevin_flansburg** (2026-06-20): https://x.com/kevin_flansburg/status/2068289253260640269 - I would strongly prefer if my doctor was just a bayesian model that ordered a series of diagnostic tests to maximize information gain given cost / invasiveness tradeoffs. This doesn’t require a human and I don’t think it even requires AI except to interface with the user. - **threepointone** (2026-06-20): https://x.com/threepointone/status/2068283951878214136 - I strongly suspect there's a straight line you can draw from SF chasing out all the artists/bands/writers/small businesses out of the city, to the lack of a clear "consumer" AI story in the world. - **DhravyaShah** (2026-06-20): https://x.com/DhravyaShah/status/2068146869533130987 - This is actually so goated... https://t.co/50iEdXYx6u - **dkundel** (2026-06-19): https://x.com/dkundel/status/2068096057310146683 - There is a UI dropdown as well but the fact that you can just tell Codex to handoff the work is incredible ❤️ https://t.co/HM7k7Lxr2n - **IgorZIJ** (2026-06-19): https://x.com/IgorZIJ/status/2068081546045575241 - Introducing the early preview of Durable Agent Sessions in https://t.co/SLQRCKIF1x 00:12 What are durable agents sessions? 02:10 Problems people hit while building background agents 03:15 Runtimes crashing/hanging 04:00 Self Healing Runtimes 04:17 Brains and Hands sandboxing 06:15 The Architecture 06:58 Sessions 08:10 Secrets 09:23 Live Streaming 11:04 Steering 12:48 Example Get an API key here: https://t.co/3PneryGXPS Demo Repo from the video here: https://t.co/CQgLCBNxSn Go build an agent, or better yet, ask your agent to! - **eastdakota** (2026-06-19): https://x.com/eastdakota/status/2068077196573352191 - Yup. https://t.co/oOjjqzXPpa - **samlambert** (2026-06-19): https://x.com/samlambert/status/2068075128723038571 - One company serves 23% of the internet and the other one hosts some of the world's largest relational databases. Good partnership. https://t.co/BeC6FdjAOK - **flueai** (2026-06-19): https://x.com/flueai/status/2068065254312128974 - Built with Flue 🦾 https://t.co/fA7vrv9OlV - **kyliestew** (2026-06-19): https://x.com/kyliestew/status/2068064544547840270 - i’m working on a benchmark to answer “which models perform best deploying cloudflare applications with docs context?” https://t.co/Ih7ApyCfoG - **clairevo** (2026-06-19): https://x.com/clairevo/status/2068056311930224781 - After dozens of interviews, I’ve found two types in companies: - super ahead on AI, feels like they are behind - pretty basic AI usage, feels like they have it all figured out - **flueai** (2026-06-19): https://x.com/flueai/status/2068044742341140667 - NEW: Evals Model output can change over time. Evals give your Flue agents a repeatable set of scenarios and expectations, so you can catch regressions before they ship. Powered by vitest & vitest-evals, a testing library by @zeeg & the @sentry crew. https://t.co/DfJ5HR9Srj https://t.co/zPra8R3XYy - **rahulgs** (2026-06-19): https://x.com/rahulgs/status/2068007838442479988 - inspect is now our digital coworker. 75%+ code at ramp now comes from inspect. what we've invested in: 1. repo setup across every main repo: is every dep installed, every command able to run and is performant, every skill token efficient and installed. are all the tools available - are the unused tools removed? which mcps are bloated? what can we precompute and snapshot in the sandbox fs (mypy, uv deps, bytecode caching, git clone). every additional source of feedback improved the final PR quality: for infra we added terraform plans, for web we added a parallel browser testing agent, for backend we added mypy's new experimental parallel mode, fast enough to run in a sandbox 2. ui polish and closing the loop outside of github - human reviews, ai reviews, diffs, child sessions, stats, fast voice inputs with realtime gpt 3. insane performance obsession (mostly @_dylanga), chasing down every ms across sandbox boot, new prompts, db queries, network round trips 4. try to match or exceed local agent performance across all axes (tools, skills, performance, repo set up) 5. robust api - used for many other automations internally 6. cost optimization: tool bloat, flex api, picking the right model, reasoning level across models 7. making sure prompts are always declarative, and almost never imperative. fix this ✅ implement this feature ✅ use the datadog mcp to fix this ❌ use the db migration skill from this link ❌ it's a moving target but a unending emphasis on speed, defaults, repo setup has allowed us to get as close as we can to "just works" on any knowledge work tasks at ramp - eng, product, data, support, sales, uxr - **mehulmpt** (2026-06-19): https://x.com/mehulmpt/status/2067976828158120006 - This is amazing, I kinda expected this after the projectsdev initiative by Stripe, but controlling abuse here is the whole game (which I believe cloudflare can do) https://t.co/UXeLQJHHYv - **threepointone** (2026-06-19): https://x.com/threepointone/status/2067970619929510350 - is there interest in a 4k+ word deep dive in building reliable agent loops (on cloudflare and elsewhere) writing down what I've done for building agents resilient to catastrophic failures on clients/servers/inference (with zero user code) and I need to get it out of my brain - **burcs** (2026-06-19): https://x.com/burcs/status/2067965146580803799 - a cloud for agents and humans https://t.co/qfUJhZy4dc - **SidChatterjee** (2026-06-19): https://x.com/SidChatterjee/status/2067957488637321665 - One small step for humans, a big step for agentkind We've been working on making it much easier for agents to deploy to Cloudflare Workers AI agents can now deploy a real app without you signing up first, doing OAuth, creating an API token, or clicking through a dashboard Just wrangler deploy --temporary https://t.co/4oKMnHvloU - **Cloudflare** (2026-06-19): https://x.com/Cloudflare/status/2067956828290302374 - Today we're rolling out Temporary Accounts on Cloudflare Workers. Any agent can now run wrangler deploy — temporary and get a live Worker in seconds. https://t.co/o5GLomVUxb - **irvinebroque** (2026-06-19): https://x.com/irvinebroque/status/2067956771847844104 - agents can now create temporary Cloudflare accounts let your agent deploy to Cloudflare without signing up or logging in wrangler deploy --temporary https://t.co/pAa1OBvHpv https://t.co/o24s4r9sdn - **LakeAustinBlvd** (2026-06-19): https://x.com/LakeAustinBlvd/status/2067940146922594531 - One of my closest friends, someone I lived with for years, is the author of this Sunday’s Modern Love article in the NYTimes. If you have a moment, his post about fatherhood will make you happy, then very sad, and then hopeful. He’s an even better person in real life than this post makes him seem. I hope we can all have his grace someday. - **mitsuhiko** (2026-06-19): https://x.com/mitsuhiko/status/2067937202596622646 - It's so cool seeing people build stuff on pi. Makes me want to do scheduled office hours. - **pheoru** (2026-06-19): https://x.com/pheoru/status/2067935755444363292 - Wouldn't it be great if you could manage all your networking routes — Tunnel, Mesh, WAN, Transit — from a single UI with an interactive network map? Well, now you can: https://t.co/TJqvRt7bMv https://t.co/gDyIzayND3 - **windymelt** (2026-06-19): https://x.com/windymelt/status/2067872723880362190 - この記事を読んで、コンパイラすらもCloudflare Containerで動作させればScala動かせるじゃんと気付いたので実際にやってみた。CFでLLMを動かしてコード生成させ、CFでコンテナを走らせてJSにトランスパイルし、CFでDynamic Workerを走らせて隔離環境で動作させてます https://t.co/bQ9VpS0UTK https://t.co/zkPH0FPiKH https://t.co/n0yOvTL2Z8 - **yusukebe** (2026-06-19): https://x.com/yusukebe/status/2067858356904357944 - `hono`のダウンロード数が2億/月近くなるんだけど、でかめのプロジェクトに依存されててありがたい 例: Flue / Mastra / MCP SDK (modelcontextprotocol/hono) / Cloudflare products / あとVoidも使ってる https://t.co/scFl4qczOY - **samlambert** (2026-06-19): https://x.com/samlambert/status/2067852239172194756 - I genuinely think CEOs should just aim to be the best collaborators. Work with people and maximize their ideas. - **TomokiSaito0920** (2026-06-19): https://x.com/TomokiSaito0920/status/2067808504287826337 - Cloudflare で AI エージェントなどの基盤を作るの、めっちゃおすすめです！ この記事の内容とは少し逸れるかもしれないのですが、最近 Cloudflare Workers のおかげでめっちゃ助かった事がありました。 例えば AI エージェントでの自動化を図る際、連携させたいメジャーなサービス (Zoom Webhook や Slack App など) で 「リクエストを受け取ってから 3 秒以内にレスポンスを返す」 みたいな必要性に迫られることがあります。 自分は「Cloud Run で FastAPI のコンテナを動かす」という構成が好きでよく使っているのですが、リクエストベースの課金にしていると費用がかなり抑えられる一方、コールドスタートのせいでこの制約に対応できないことがあります。 （単に Cloud Run の CPU を常時割り当てておけば解決するのですが、そのサービスのためだけに月 2,000 円とかかかるのは嫌です。） そこで、前段に Cloudflare Workers で 「イベントは署名検証して Cloudflare Queues に積んで即 200 を返す」 （Queue から Cloud Run に通知を出して、処理そのものはのんびり行う） という対策をしました。 「状況は結局変わらないんじゃ？」と思うかもしれないのですが、コンテナではなく V8 Isolate のおかげでこの 3 秒ルールをだいぶ余裕を持ってクリアできました！ GCP とか AWS, Azure であれこれ頑張るという道もあるのですが、費用がかかったり構成が複雑になったりしてしまうでしょう。 Cloudflare Workers の Workers Paid Plan だと $5.5 / month だけ（しかもこの課金だけでかなりたくさん使える）でシンプルに解決できるということが多いですし、他にも Workers AI などのかなりイケてるサービスが多くあります！ 「これは AI エージェントや自動化の基盤を作るときにめちゃくちゃ便利だなぁ」 と思いました🔥 - **yusukebe** (2026-06-19): https://x.com/yusukebe/status/2067803970547114226 - まじでエージェントやらないと取り残される。ハーネスエンジニアリングでAGENTS.mdを整備しましょう、リンター用意しましょうとかそういうのじゃなくて - **yusukebe** (2026-06-19): https://x.com/yusukebe/status/2067800184353083457 - Flue面白い みんなでやろう https://t.co/MyoSm7m3hZ - **yusukebe** (2026-06-19): https://x.com/yusukebe/status/2067799548223963219 - Fred, you did well! Flue is built on Hono, and we can use Hono in the Flue app. I got it now https://t.co/gdemlAyDgU https://t.co/5MbkvrLvS1 - **BarrettsGoat** (2026-06-19): https://x.com/BarrettsGoat/status/2067798088262062245 - This genuinely one of the coldest pics ever holy shit https://t.co/cV8joKEzXq - **Coopz___** (2026-06-19): https://x.com/Coopz___/status/2067776288748646854 - i think people are not conceptualizing the historical significance of this video https://t.co/dmyHs3Fp7A - **CFchangelog** (2026-06-19): https://x.com/CFchangelog/status/2067768202289901998 - Create PlanetScale Postgres and MySQL databases directly from Cloudflare. Use them with Workers and Hyperdrive for connection pooling and caching. https://t.co/nyxWTacWSL - **boymolish** (2026-06-18): https://x.com/boymolish/status/2067738626222346271 - This shot by Mitch Ritter😮‍💨 https://t.co/tYeTXBD83Z - **Dylan_Burd** (2026-06-18): https://x.com/Dylan_Burd/status/2067710257950863580 - My favorite video I’ve ever taken. Absolutely surreal. https://t.co/XP5UVrvtq8 - **FredKSchott** (2026-06-18): https://x.com/FredKSchott/status/2067702675622465559 - fun fact: it took astro 2 years to hit these numbers https://t.co/GjWSjKQU9H - **gaganbiyani** (2026-06-18): https://x.com/gaganbiyani/status/2067687179917373466 - The current AI wave is the first generation - deeply technical, setting up the foundation. Mostly founded by scientists and engineers. Similar to Apple, Microsoft or Google. Then the application layer will get in full swing (it’s already here but it’s gonna get so much bigger). The AI era’s Amazon, Salesforce, Airbnb, etc. The pattern repeats every wave: infrastructure first (engineers), then applications that touch millions of users (design, product, + sales/marketing leaders). - **ritakozlov** (2026-06-18): https://x.com/ritakozlov/status/2067611714288332986 - who's going to @aiDotEngineer in sf coming up? come say hi to the cloudflare team at the expo, or at coffee at sightglass on tues/ run club on thurs PS: first 50 attendees for run club get 404 running shirts ☕️ -> https://t.co/xmN92ufKXh 🏃‍♀️ -> https://t.co/VGx8p6nKQG - **boshen_c** (2026-06-18): https://x.com/boshen_c/status/2067454889257840844 - I started patching the rust port of the react compiler to make it faster. https://t.co/hrpCptRVUJ - **sama** (2026-06-18): https://x.com/sama/status/2067427678529974740 - We offer no explanation as to why Noams are so good at AI; we attribute their success, as all else, to divine benevolence. https://t.co/QQ93UCF4ok - **RhysSullivan** (2026-06-17): https://x.com/RhysSullivan/status/2067387191714812112 - there's this valley of despair when programming with agents you either have to be an inference maximalist or minimalist, but anywhere in between you just get slop and pain https://t.co/JhnACKYzrH - **zeeg** (2026-06-17): https://x.com/zeeg/status/2067386142375514354 - https://t.co/8LCL0qhALl - **FredKSchott** (2026-06-17): https://x.com/FredKSchott/status/2067341484731568423 - @yusukebe https://t.co/El9dtemsAO - **thomasgauvin** (2026-06-17): https://x.com/thomasgauvin/status/2067334809869848619 - The Astro team building the framework layer for production agents makes a lot of sense. Great DX, deploy anywhere. Congrats @FredKSchott & team! And with Flue's Cloudflare target, each agent becomes a Durable Object, with durable execution/codemode/sandboxes (blog below!👇) https://t.co/zJwm9ziMxf https://t.co/21DPlZhGJA - **FredKSchott** (2026-06-17): https://x.com/FredKSchott/status/2067302366718947778 - some thoughts on eve, especially re: @flueai and the larger agent builder space. The biggest surprise: eve is vercel-only. You can't deploy it outside of vercel and it requires a full bet on their stack -- vercel runtime, vercel sandbox, vercel workflows, vercel cron jobs, etc. etc. short-term limitation, or the start of a larger "managed agents" play by vercel? eve is filesystemmaxxing. this feels specifically optimized for agents, who love the codebase consistency and have no trouble keeping track of 10+ magic file and folder conventions as the project scales to 100s of files. (^^ this is probably the most interesting design decision here. I have a ton of respect for anyone taking a big swing like this if this is intentionally optimized for agents. makes total sense.) eve has a custom-built harness. probably not interesting to most people. I'm curious if they will leverage the new ai sdk HarnessAgent concept to swap different harnesses. otherwise there's a lot of great stuff here. lots of good features. The dev-server-as-TUI is great. you can tell how much attention and care went into this. excited to see our two projects continue to explore the same space from different directions! - **jpschroeder** (2026-06-17): https://x.com/jpschroeder/status/2067302005966520573 - Current stack: - Cloudflare Domains - Cloudflare Workers - Cloudflare Durable Objects - Cloudflare Artifacts - Cloudflare Workers AI - Cloudflare Queues - Cloudflare D1 - Cloudflare Containers - Cloudflare R2 - Cloudflare Workers for Platforms - **ritakozlov** (2026-06-17): https://x.com/ritakozlov/status/2067265739166187722 - flue is such an incredible all-in-one framework for building agents the built in handling for channels and durability is 🧑‍🍳😘 and can be deployed easily to cloudflare or.... literally anywhere else https://t.co/cgSAMZ6nql - **elithrar** (2026-06-17): https://x.com/elithrar/status/2067242070499058095 - 1000%. this is a huge part of our culture. artifacts started like this: @mattzcarey had a prototype and had been noodling on the idea. we were seeing existing source control services struggle to scale. I was speaking to customers and we said… what if? https://t.co/wSwQi15LLT - **ritakozlov** (2026-06-17): https://x.com/ritakozlov/status/2067240033543680433 - at a lot of companies, product's role is to come up with ideas, carefully groom the roadmap and narrowly define requirements for engineering to (blindly) follow this maybe makes for an "easier" product role but limits creativity (and accounrability) one thing that's unique about cloudflare is that ideas can really come from so many more places product's role is to help map those ideas to customer problems and make sure we actually solve them and help get those ideas in customers' hands (aka actually ship it and make it good!) it makes for a much more interesting role and breeds so much innovation and leads to better experiences because engineering is not exempt from taking ownership in the deliverable. "i shippped what's in the PRD" is not good enough. you own the customer problems & solutions together - **korinne_dev** (2026-06-17): https://x.com/korinne_dev/status/2067237787456393624 - Last day of my six weeks in Bengaluru and my team woke up at 4am to take me to Nandi Hills 🫶 very sad to leave and already thinking of when to come back. I’ll remember this trip to the @Cloudflare office for the rest of my life - thank you @KankaniAbhishek for hosting me and building such an incredible culture at the office. I’ll miss you all (and the masala dosa) - **benjaminsehl** (2026-06-17): https://x.com/benjaminsehl/status/2067190846223429937 - So excited about this! Been building something with Flue and it’s really come a long way in such a short period of time. https://t.co/F0DvFgI7rR - **yusukebe** (2026-06-17): https://x.com/yusukebe/status/2067122972364775782 - TODO: Flueで遊ぶ https://t.co/IvMXQPsUb9 - **cnakazawa** (2026-06-16): https://x.com/cnakazawa/status/2067008791040889224 - First thing I helped out with at Cloudflare ☺️ Flue is pretty good, actually. https://t.co/zMf5rzZ7WE - **matthewcp** (2026-06-16): https://x.com/matthewcp/status/2067000856797700570 - We use Flue in Astro to run issue triage in GH CI. It reproduces the bug report, diagnoses the cause, verifies that it is a bug (and not intentional behavior) applies a fix, writes a test, and then creates a preview release for users to try. Here's an end-to-end fix in about an hour: https://t.co/v4YRftLCLr - **mitsuhiko** (2026-06-16): https://x.com/mitsuhiko/status/2066985155332280677 - Powered by Pi :) https://t.co/dhFrKZJpeh - **dok2001** (2026-06-16): https://x.com/dok2001/status/2066980285393031632 - .@flueframework just shipped 1.0 Beta. Durable agents, any LLM, deploy anywhere. Built by the @astrodotbuild team on Durable Streams, Vite, and a flexible Sandbox API. https://t.co/pquje18GIX - **ankrgyl** (2026-06-16): https://x.com/ankrgyl/status/2066978626822865279 - Very excited about Flue and the new native integration with Braintrust. Agent tracing is quite a bit different than traditional LLM apps and evolving quickly. https://t.co/pyS0uk7tWb - **elithrar** (2026-06-16): https://x.com/elithrar/status/2066971518983098514 - literally porting the code review agent we use across all of our OSS repos to Flue as we speak. https://t.co/6EI1BwsFmL - **dillon_mulroy** (2026-06-16): https://x.com/dillon_mulroy/status/2066965896388604244 - flue. the open agent framework. i’ve been building a bunch of my agentic workflows on the primitives in the 1.0 beta - now you can too https://t.co/t25fcGrL4V - **badlogicgames** (2026-06-16): https://x.com/badlogicgames/status/2066965697016561811 - approved https://t.co/rv3sc2u2xo - **zeeg** (2026-06-16): https://x.com/zeeg/status/2066964926137241876 - ✅ bugs go into Sentry we're exploring Flue for some of our CI workflows https://t.co/rlAXicNxTJ - **FredKSchott** (2026-06-16): https://x.com/FredKSchott/status/2066962296119959581 - Just Shipped: Flue 1.0 Beta Flue is the TypeScript framework for building the next generation of agents, designed around an open agent harness with zero LLM lock-in. It’s like Astro, for agents. Flue 1.0 has been redesigned around three core primitives: 🔁 Workflows — structured automations designed for background work, where your code drives the agent from start to finish. 🧭 Agents (New!) — autonomous, stateful loops where the model drives itself to complete a given task. 📡 Channels (New!) — connect agents to Slack, GitHub, Linear, Discord, Teams, and more. Flue handles the boilerplate for you. Everything shares the same durable foundation, powered internally by Pi, Vite, and Durable Streams. Deploy anywhere, use any LLM, and recover running agents across restarts and downtime. We’ve talked to a lot of teams building agents, and keep hearing the same thing: getting to production is hard work. We built Flue to help change that. Flue 1.0 Beta is available today. Give it a try and let me know what you think! - **dillon_mulroy** (2026-06-16): https://x.com/dillon_mulroy/status/2066921030346535096 - FINALLY!!!!!!!!!! https://t.co/UaJL2jWED6 - **overlook_motel** (2026-06-16): https://x.com/overlook_motel/status/2066853669543309791 - This weekend I was shown something astonishing. Oxc's parser is already the fastest JS/TS parser available. Now someone has found a way to make it even faster - by up to 3x! A tour-de-force of engineering, based on a truly novel approach. 🤯 Watch this space... - **thdxr** (2026-06-16): https://x.com/thdxr/status/2066764365819261083 - i hope it's clear now why open source models are important i've said before i can respect the position around safety but it's completely naive even if you think you have superior morality and should control it someone will kick you out and take control - **citrini** (2026-06-16): https://x.com/citrini/status/2066712653297299797 - AI workloads are creating new memory objects that are too large for HBM but too valuable for cold storage. It’s simply to expensive to use DRAM for everything if it’s possible to use NAND for some of it. We’ve gotten three announcements focused on flash from three major players in the past couple weeks: Nvidia is using flash to store reusable KV cache via CMX. This targets long-context, multi-turn and agentic inference by extending GPU memory with a shared KV-cache tier, allowing flash to become context memory for long-running agents. AMD is using flash to store some system memory until it is predicted to be hot enough to go to DRAM. They’re buying MEXT as a software layer to accomplish this. Apple is using flash to store some model weights until needed. Apple stores inactive experts in flash on the device side (AFM 3 Core Advanced, published a week ago). If you can get the system to know what data is likely to be needed next, flash goes from simple storage to cheaper memory. - **zeeg** (2026-06-15): https://x.com/zeeg/status/2066670670893965731 - Who are my SF tech company peers that are Valkyries fans? I need some coconspirators 👀👀 - **dillon_mulroy** (2026-06-15): https://x.com/dillon_mulroy/status/2066552096325865886 - can’t wait for zojirushi code https://t.co/x01fKXqPdg - **michellechen** (2026-06-15): https://x.com/michellechen/status/2066527619999264907 - welcoming ensemble ai to cloudflare. don't mind if our models get a little faster. we're deepening our investment in this space and our team keeps growing, you should join us https://t.co/aVNcIwDkDK - **satyanadella** (2026-06-14): https://x.com/satyanadella/status/2066182223213293753 - https://t.co/vLmiBKTtX3 - **kristianfreeman** (2026-06-14): https://x.com/kristianfreeman/status/2066174559678554568 - The legacy of this Finals for the Spurs - Harper was the best player on the team. It was brutal to see the Spurs not adjust... ever. Kept running plays for Fox over and over. Excruciating to watch. If you're a Spurs fan you're spending this summer feeling really unsettled https://t.co/hVGolpJaBy - **eastdakota** (2026-06-14): https://x.com/eastdakota/status/2066171789160464579 - This is a bigger trend. Cracks forming fast. Lots of schools are going to fail. https://t.co/RkvT4ZVANc - **abditmr** (2026-06-14): https://x.com/abditmr/status/2066006705100484930 - can’t stop thinking about this bucket https://t.co/LJwBk9Ft9Z - **jaycaspiankang** (2026-06-14): https://x.com/jaycaspiankang/status/2066002781115396155 - Wemby went from esoteric dude who trained w the monks and did things we had never seen before on a basketball court to ‘FUCK THAT FRENCH BITCH’ in the course of a month. It’s what makes America great lol. - **tatefrazier** (2026-06-14): https://x.com/tatefrazier/status/2066002778976305514 - Nike still nails these commercials every time https://t.co/EZeIT79OM3 - **maxkellerman** (2026-06-14): https://x.com/maxkellerman/status/2066002402055213448 - Still want Wemby suspended for Game 6 - **FOS** (2026-06-14): https://x.com/FOS/status/2066001292175024564 - Nike’s ad for the Knicks winning their first championship since 1973. “Never slept. Always dreamed.” https://t.co/bMF2vj8sKp - **andre** (2026-06-14): https://x.com/andre/status/2066000800938360890 - MIKE BROWN!!! - **NYCSanitation** (2026-06-14): https://x.com/NYCSanitation/status/2066000135943770239 - NEW YORK CITY YOU LISTEN TO ME! IF YOU’RE NEAR SOME TRASH RIGHT NOW, ANY TYPE OF 24-HOUR TRASH, GO DISPOSE OF THE TRASH RIGHT NOW AND PUT YOUR HAND INSIDE THE OFFICIAL NYC BIN! YOUR TRASH IS OUR TRASH AS OF RIGHT NOW. CONGRATS KNICKS 💙🧡💙🧡💙🧡💙🧡💙🧡💙🧡 - **BillSimmons** (2026-06-14): https://x.com/BillSimmons/status/2065995176041218052 - This is an unfathomable Brunson performance even for him. - **StephanieWei** (2026-06-14): https://x.com/StephanieWei/status/2065972801924747485 - Dude just showed up handing out pizzas, people cheering each other on to close down the street, everyone scooching over to make room for one more … never seen more love and unity. #knicksinfive https://t.co/nAfEVXMVW6 - **ShabazzStuart** (2026-06-13): https://x.com/ShabazzStuart/status/2065857805475553565 - I feel like the fact that Jalen's dad played for the Knicks in the finals against the Spurs back in 1999 when Jalen was just two years old is legitimately under-appreciated. What a dope NYC story. https://t.co/iO6nGb3BdS - **roerohan** (2026-06-13): https://x.com/roerohan/status/2065792842929476038 - We just wrapped up our @CloudflareDev buildathon and it was a full-house! Thank you for showing up and shipping with us! We should do this more often, shouldn't we? 👀 Shoutout to @cengines_studio for letting us use their beautiful office https://t.co/A0wUd3JDJz - **dillon_mulroy** (2026-06-12): https://x.com/dillon_mulroy/status/2065581796520063269 - we got another one folks @0xblacklight is now DO pilled https://t.co/3rupBc79lT - **elithrar** (2026-06-12): https://x.com/elithrar/status/2065552585667309938 - working on an specialized agent and wanted to add an invisible @cloudflare Turnstile to tackle any bots. one of our interns (Jules!) just shipped Turnstile Spin, which is a guide for agents to follow + set up Turnstile on any site: https://t.co/8hcluDG87f perfect timing. https://t.co/zbfHo93orX - **eersnington** (2026-06-12): https://x.com/eersnington/status/2065550037183279543 - a side effect is any operation that changes or depends on state outside the workflow, like charging stripe, sending an email, writing to a database, or calling a webhook, yada yada. with stepdaddy you explictly define the side effect once, including how it should recover. then inside a workflow step you call it with the workflow context + a stable key. so when the step retries, stepdaddy knows whether to reuse the saved result, safely call the API again with the same idempotency key, or stop before doing something stupid twice - **RhysSullivan** (2026-06-12): https://x.com/RhysSullivan/status/2065540963074805982 - A stupidly high leverage move for companies right now is to vibe code a test environment for people to use anonymously with your product Throw it on Cloudflare w/ Durable Objects for state, enable people to use your product in their flows realistically with no API keys set - **elithrar** (2026-06-12): https://x.com/elithrar/status/2065532307075650021 - we're hiring for one of the fastest growing and most impactful roles @cloudflare right now. R2 and its storage infrastructure underpins a huge array of our products and platforms. https://t.co/oMlIag4M2Q - **acoyfellow** (2026-06-12): https://x.com/acoyfellow/status/2065512376980447390 - Serve an OpenTUI application over real SSH from a @Cloudflare Container > https://t.co/IftGq20b5L > https://t.co/oX0yJn6UXh - **roerohan** (2026-06-12): https://x.com/roerohan/status/2065496056583123045 - @michellechen crazy how I found out about the existence of kimi k2.7 from the post about it's availability on workers AI before the launch post - **michellechen** (2026-06-12): https://x.com/michellechen/status/2065494981423362132 - k2.7 code live on workers ai https://t.co/QLzuPfJymT - **kevin_flansburg** (2026-06-12): https://x.com/kevin_flansburg/status/2065494855229272142 - Another day, another day 0 launch on Workers AI. This time @Kimi_Moonshot's Kimi K2.7 Code, focused on long horizon agentic coding tasks, and featuring 30% improvement in reasoning efficiency over Kimi K2.6! https://t.co/xS50gtw4bJ - **ritakozlov** (2026-06-12): https://x.com/ritakozlov/status/2065432781669245311 - the biggest voice / video chat app you know runs on @cloudflare compute https://t.co/NpyhLFVbnR - **roerohan** (2026-06-12): https://x.com/roerohan/status/2065372578718650825 - Flagship now has API support. You can now interact with @CloudflareDev Flagship programmatically - enabling workflows, automations, integrations, and agent-driven usage beyond the UI. 🧵 https://t.co/w3nuHN6Z7V - **RhysSullivan** (2026-06-11): https://x.com/RhysSullivan/status/2064939888584384698 - @jlongster admittedly i haven't dove into the spec behind them, but i just sort of view them as a unified way to get an iframe rendering in a client in that iframe you can do whatever you want, i.e this one of generative ui https://t.co/0LPDyeoRhP - **miawang__** (2026-06-11): https://x.com/miawang__/status/2064933211331777022 - Mood: https://t.co/O0VJnNa0dv - **jerryferrara** (2026-06-11): https://x.com/jerryferrara/status/2064924211081076998 - KAT truly may have saved the day. https://t.co/4dlf9hN68v - **NIKNBAYT** (2026-06-11): https://x.com/NIKNBAYT/status/2064919717152739646 - Karl-Anthony Towns saved the whole comeback. I can only respect it and shake his hand https://t.co/ieLNqtSO0M - **dillon_mulroy** (2026-06-10): https://x.com/dillon_mulroy/status/2064825204241543196 - @CodeWithZeee cc @irvinebroque - **csparks1919** (2026-06-10): https://x.com/csparks1919/status/2064773495607980205 - @dexhorthy prop drilling is such a consistent problem. fable seems a bit better at taste tho been meaning to try this out, but I wonder if adding a ton of bespoke linters to limit prop number or types would mitigate the issue. i've tried a similar thing for test writing, and it's worked - **SidChatterjee** (2026-06-10): https://x.com/SidChatterjee/status/2064676982609051750 - My mom is 70 and had never worked out until a year ago when I encouraged (read: forced) her to start with a personal trainer to protect her joints and muscle as she’s getting older I’m so proud of her, she’s stronger, fitter, has zero joint trouble anymore and is loving it https://t.co/8PGtnKXgXQ - **0xblacklight** (2026-06-10): https://x.com/0xblacklight/status/2064553409395724631 - I interviewed close to two dozen people this week and something I heard a lot of is "I don't think about the code too much but I think a lot about system design and architecture" I don't think that's quite right, and here's why: before you ever get to system design you should think about program design system design is important! it matters a lot for scalability. but if you don't think about your type system and if you don't carefully design your seams and figure out how to make your code testable (you should probably use dependency injection btw) and if you don't think about where state lives and how it's managed and if you don't think about control flow and where abstractions should and should not exist your code is going to be an unmaintainable, poorly-factored mess of bad types and spaghetti code and even minor changes will turn into shotgun surgery and MASSIVE diffs I have seen it done I have even done it myself and it has never ended well "GPT-7 will fix it" does not help you when there's an incident at 3am that the agents can't debug and now you can't debug it either and now you have to unwind months of bad code I also have never heard "I don't look at the code, just at the system design" said by someone who is actually good at system design program design and system design are more closely coupled than people think this was a very strong (negative) predictor of how someone would do on the system design part of the interview make of this what you will - **michellechen** (2026-06-10): https://x.com/michellechen/status/2064499013370446032 - fable 5 from @AnthropicAI now on @Cloudflare AI Gateway https://t.co/lLTlWKNGm9 - **dillon_mulroy** (2026-06-09): https://x.com/dillon_mulroy/status/2064470331822055590 - artifacts https://t.co/IbJxzaCyAB - **zeeg** (2026-06-09): https://x.com/zeeg/status/2064459353617473602 - just pushed our new search/execute primitives to non-experimental in MCP the tool surface is now both smaller, and larger let me know how it works for you! - **ritakozlov** (2026-06-09): https://x.com/ritakozlov/status/2064414840391704830 - code mode has helped @cloudflare save ~93% on token cost talking to peers in the industry whose spend is creeping towards upwards of $100k per engineer per year you do the math on the cost savings of that for a large engineering org https://t.co/1qSGOB9vuT - **astrodotbuild** (2026-06-09): https://x.com/astrodotbuild/status/2064383758082458002 - @RobPando - Updated to Vite 8 - Rewritten Astro compiler in Rust - New Rust-powered Markdown pipeline by default TL;DR — faster builds across the board Our draft migration guide has some details on breaking changes: https://t.co/119ykm9T1y - **destraynor** (2026-06-09): https://x.com/destraynor/status/2064382547430195481 - Excited to say that @hubspot's 300,000 customers and @freshdesk's 150,000 customers can now use @fin_ai to power their Customer Service, giving their customer's instant accurate excellent answers. Links in the thread. https://t.co/r3YuT9QjFt - **mattzcarey** (2026-06-09): https://x.com/mattzcarey/status/2064377811180093713 - in the past few days the Cloudflare MCP server made 2.6M API requests from 735k agent tool calls. code mode enables agents to do more per tool call. - **astrodotbuild** (2026-06-09): https://x.com/astrodotbuild/status/2064364152861130754 - Shh, don’t tell anyone, but… https://t.co/D0IqKxElSm - **dexhorthy** (2026-06-09): https://x.com/dexhorthy/status/2064351095632380412 - I have some thoughts on this but you will have to come to the coding agents track at @aiDotEngineer worlds fair if you want to hear them :) https://t.co/dINW9kxl26 - **acoyfellow** (2026-06-09): https://x.com/acoyfellow/status/2064305599614001183 - today everyone's discussing agent loops, costs, tokenmaxxing.. The common assumption is that only big token companies with unlimited budgets can sustain them. But @Cloudflare enables cost-controlled loops for anyone: Here is an outer cron spawns dynamic inner workflows, imagine it paired with AI Gateway spend limits... loopmaxxing, at your own speed. https://t.co/xhlwVDwoeY - **acoyfellow** (2026-06-09): https://x.com/acoyfellow/status/2064304074518921327 - @nurodev Oops, it got flatter https://t.co/eSkLx7jaSQ - **nurodev** (2026-06-09): https://x.com/nurodev/status/2064303188035956826 - @acoyfellow Damn, not using the newest latest & greatest workflows features? 😉 https://t.co/nXpWTWS5Jx - **CFchangelog** (2026-06-09): https://x.com/CFchangelog/status/2064287790922109392 - SMTP submission is now available in beta for Cloudflare Email Service. Send emails through standard clients like Nodemailer and smtplib. https://t.co/uUD4jxzcJs - **RhysSullivan** (2026-06-09): https://x.com/RhysSullivan/status/2064234368143691818 - good new detail from cloudflare, after signing in to wrangler CLI they help you get setup with an agent https://t.co/KTsoMP7Gmk - **samlambert** (2026-06-09): https://x.com/samlambert/status/2064194313677127730 - 12/31/1999 https://t.co/f8tp2jpR4C - **zeeg** (2026-06-08): https://x.com/zeeg/status/2064089493435744446 - benchmark that shows every model is mediocre af writing worthwhile code i like it, and it represents something closer to the truth https://t.co/UBQsuIWbxo - **sama** (2026-06-08): https://x.com/sama/status/2064088940932641225 - Here is our current plan for OpenAI: https://t.co/r29FUUee3A - **lessin** (2026-06-08): https://x.com/lessin/status/2064079288702570647 - https://t.co/yPUoQ8v2WC - **threepointone** (2026-06-08): https://x.com/threepointone/status/2064074704365686821 - @roerohan @boristane sure they can run event concurrentlys? like requests etc. I think the alarm should trigger even if an http request is being worked on - **dillon_mulroy** (2026-06-08): https://x.com/dillon_mulroy/status/2063982435474252077 - i might be cooking https://t.co/fHbqatFSJY https://t.co/EnlUwbaBSQ - **elithrar** (2026-06-07): https://x.com/elithrar/status/2063727073072443682 - a founder just messaged me the other day telling me they wanted to get their customer config & workflow definitions into Artifacts so that it’s versioned by default. store a commit hash in your DB as a pointer to the latest. let us take care of the rest. - **fayazara** (2026-06-06): https://x.com/fayazara/status/2063338457389502851 - tons of updates in Screendrop v0.20 https://t.co/nX3zarcx4z - **dok2001** (2026-06-06): https://x.com/dok2001/status/2063317783333536138 - The best founders play the long game and willing to take educated bets. Something I have gotten to experience with @eastdakota https://t.co/SMq11SAVWi - **CAPTAINPEARLNYK** (2026-06-06): https://x.com/CAPTAINPEARLNYK/status/2063248123628024309 - This is the single greatest postgame interview after a WIN that I've seen and every youth player needs to watch. Passion. Accountability. Humility. Go Knicks https://t.co/CmGR85zJGc - **RhysSullivan** (2026-06-06): https://x.com/RhysSullivan/status/2063139769719402850 - dumb observation, but LLMs lack the ability to reason from another persons perspective what a task would be like example of this is I’m trying to have it build a test suite that feels like a real person using the product, and it can’t conceptualize it - **FredKSchott** (2026-06-06): https://x.com/FredKSchott/status/2063123765219963147 - 🤖🇯🇵 https://t.co/asJKC7VLmw - **awealthofcs** (2026-06-06): https://x.com/awealthofcs/status/2063100147723124782 - Wemby is like AI's impact on the labor market It feels inevitable but it's not happening quite yet - **dillon_mulroy** (2026-06-06): https://x.com/dillon_mulroy/status/2063064188386210273 - artifacts https://t.co/LeAOwnT6uP - **andrewingram** (2026-06-06): https://x.com/andrewingram/status/2063049889861316649 - @ElectricSQL Maybe I’ve misunderstood, but Cloudflare’s Project Think isn’t like the others, it doesn’t run inside a sandbox. It’s closer to the architecture you’re advocating. - **threepointone** (2026-06-05): https://x.com/threepointone/status/2063029065582350402 - depending on your usage, cloudflare agents just got muchhhhh cheaper. update to latest please. (thanks @FredKSchott for the nerd snipe) https://t.co/WY5xIZ4rAV - **dok2001** (2026-06-05): https://x.com/dok2001/status/2063026091984367674 - @brendanrc2 We are! We should show you the architecture we are using! - **mattzcarey** (2026-06-05): https://x.com/mattzcarey/status/2062979141087957131 - cloudflare mcp server is cranking https://t.co/cqyI1995tK - **aninibread** (2026-06-05): https://x.com/aninibread/status/2062960132917530984 - All you need is in AI Gateway. @minglu and team have been killing it!! https://t.co/d0soV1Nezp https://t.co/wzfRX0VeYr - **dinasaur_404** (2026-06-05): https://x.com/dinasaur_404/status/2062948983983452410 - version skew protection coming to Workers! 🙌 want to know more? go talk to Angus :) https://t.co/xNWIbwyyxv - **juecd__** (2026-06-05): https://x.com/juecd__/status/2062910591287828681 - Firecracker VMs only gets you part of the way to sub-second performance. We've been further optimizing everything around it to get to the <20ms latency we're known for: minimizing disk i/o bottlenecks, auto-resizing pools to maximize hot hits & more. @sjmiller609 cooks. https://t.co/uZkS0Pmfsc - **sjmiller609** (2026-06-05): https://x.com/sjmiller609/status/2062896795626201123 - The key to ultra low latency provisioning is resource pooling. But when it comes to scale, the key is fast cold starts. Here’s the math we use to place 1st on every performance benchmark for browsers. 🧵1/5 https://t.co/0PlFcRZb3H - **michellechen** (2026-06-05): https://x.com/michellechen/status/2062894017545720129 - ai gateway cost controls, now live. what i'm particularly excited about is what's next: integrating with Cloudflare Access so you can see usage and set limits based on IdP resources (user, group, service, etc). unique to us because we have an entire ZT+ dev platform https://t.co/fHOcAiLFuh - **elithrar** (2026-06-05): https://x.com/elithrar/status/2062887228909527346 - We're seeing teams reckoning with their AI costs and wanting more control, so we shipped it. AI Gateway now has spending limits, fallbacks to more cost-efficient models when you hit those limits, and (soon) user & group based controls via your IdP + Cloudflare Access. https://t.co/YDww0mkVCK - **thdxr** (2026-06-05): https://x.com/thdxr/status/2062886815598592085 - even if i'm not actively shipping code i'm constantly talking to opencode to think through ideas you can see on days where i'm busy with company stuff my token count goes down so unfortunately this is a good measure of something - **boshen_c** (2026-06-05): https://x.com/boshen_c/status/2062849192637432159 - TIL about https://t.co/bjawZ9JwtU, an agent harness framework being developed by the astro team. - **astrodotbuild** (2026-06-05): https://x.com/astrodotbuild/status/2062843655338037679 - Welcome to the team @ocavue https://t.co/utnAdu5EaV - **0u0i3** (2026-06-05): https://x.com/0u0i3/status/2062841693054148788 - Astroチームが作ったTypeScript製エージェントフレームワーク「Flue」が面白い。Claude Codeみたいな自律エージェントを作るための環境をコードで定義できるらしい。 https://t.co/M6vdy4TH6m - **typesfast** (2026-06-05): https://x.com/typesfast/status/2062791307094048937 - During a pitch a prominent VC once told us the market size for global logistics was only $6B. My CFO’s response: “So you’re saying it’s smaller than the market for USB cables?” https://t.co/us4qby1cLl - **theo** (2026-06-05): https://x.com/theo/status/2062741740910702764 - There is a standard. It's Agents.md. Anthropic refuses to use the standard. https://t.co/rnEP5Vu3l1 - **RhysSullivan** (2026-06-05): https://x.com/RhysSullivan/status/2062718889361039812 - who's ready for recursive self improvement summer - **thdxr** (2026-06-05): https://x.com/thdxr/status/2062710020375011436 - how to be good at your job - realize this one thing is actually made up of two separate things - realize instead of solving the direct problem you can solve a broader problem - instead of implementing thing, implement other thing that makes it easier to implement thing - **sapphi_red** (2026-06-04): https://x.com/sapphi_red/status/2062538393033273676 - I'm excited to join Cloudflare! 😀 I'll continue working on Vite and Rolldown just as before, and Vite 8.1 is already on its way with some interesting new features 👀 - **evanyou** (2026-06-04): https://x.com/evanyou/status/2062533668233756677 - I've left most of what I want to say in the VoidZero blog post. But worth repeating: Thank you @voidzerodev team for trusting me and joining me on this wild ride. I am very proud to have assembled such a talented team and even prouder of what we have built together. Thank you all our investors for believing in my vision, in particular @caseyaylward from @Accel who led both our Seed and Series A. Thank you the @vite_js community. Vite and VoidZero wouldn’t have come this far without your trust and support. We will continue building with all of you, together, in the open. And thank you to everyone that made this happen at @Cloudflare. Looking forward to working with you all! https://t.co/0ly53VCOSr - **dok2001** (2026-06-04): https://x.com/dok2001/status/2062526085086990390 - Few people have shaped how the web is built more than @evanyou and the VoidZero team. Vite. Vitest. Rolldown. Oxc. Vite+. Tools developers and agents reach for without thinking. https://t.co/6g0nTteWmP - **vite_js** (2026-06-04): https://x.com/vite_js/status/2062525206158078047 - Today, VoidZero joins Cloudflare. Vite remains MIT, vendor-neutral, and stewarded by the same wider team. The same goes for Vitest, Rolldown, and Oxc. Cloudflare is also committing $1M to an OSS fund to support independent development in the Vite ecosystem. https://t.co/8B06UjeYz3 - **eastdakota** (2026-06-04): https://x.com/eastdakota/status/2062524335475032447 - The best tools to build the agentic future: all native on @Cloudflare. Welcome to the team, VoidZero!! https://t.co/Q6rMCMVHnJ - **Cameron_C2** (2026-06-04): https://x.com/Cameron_C2/status/2062521736856248642 - We're joining cloudflare! https://t.co/f42vNmPvIh - **ritakozlov** (2026-06-04): https://x.com/ritakozlov/status/2062521339190169896 - BIG DAY! @voidzerodev is joining @cloudflare 🚀 before anything else: @vite_js is very much remaining open source, and will always remain that way, with robust ongoing investment we're also going to keep making cloudflare the best place for building applications! thrilled to have @evanyou and team join us to help us on that mission. https://t.co/OndkduVZRc - **Cloudflare** (2026-06-04): https://x.com/Cloudflare/status/2062521221132992533 - VoidZero, the team behind Vite, Vitest, Rolldown, Oxc, and Vite+, is joining Cloudflare. Vite stays open source, vendor-agnostic, and built for everyone. https://t.co/DJTpX4Q9Xt - **voidzerodev** (2026-06-04): https://x.com/voidzerodev/status/2062520542121304146 - VoidZero is joining Cloudflare. Our mission stays the same: to make JavaScript developers more productive than ever before. Vite, Vitest, Rolldown, Oxc, and Vite+ remain MIT-licensed. Evan and the VoidZero team will continue leading them. Cloudflare shares our commitment to open source. Together, we can keep investing in the tooling developers rely on every day, while bringing the Vite ecosystem and Cloudflare’s platform even closer together. - **wesbos** (2026-06-04): https://x.com/wesbos/status/2062520527151903090 - Cloudflare has acquired Vite / VoidZero Void is vite's fullstack Intertia-like framework. This gives Cloudflare control over the entire stack. They have all the primitives from frontend/backend framework, linting, testing, formatting, JS runtime, db, kv, inference, blob storage, access, etc... smart move. A tidy package they can hand to an LLM to make a site. - **kristianfreeman** (2026-06-04): https://x.com/kristianfreeman/status/2062520371618738422 - Cloudflare 🤝 @voidzerodev https://t.co/jNZEGhXeKS - **thomasgauvin** (2026-06-04): https://x.com/thomasgauvin/status/2062512156076048447 - first ship on the new team: new docs for agents on Cloudflare! 🚀 really helps to visualize the breadth that @Cloudflare agent platform has to offer: Communication channels: - chat/slack/webhook agents - voice agents with voice - email agents with email service Core agent: - think as ootb harness - bring your own harness on top of agents sdk and durable objects Tools: - sandbox for code execution or running sandbox agents - MCP (with MCP server portals to secure) - AI search for document agents - payments for agentic payment All with built-in observability - **gaganbiyani** (2026-06-04): https://x.com/gaganbiyani/status/2062412184995696652 - What's more crazy is this is how the system is designed. California INTENTIONALLY does not count ballots for weeks. And I think it explains the state's broader dysfunction. Basically, California's voting rules are designed to maximize accessibility at all costs. ALL costs. Ballots can arrive after Election Day. Signatures can be fixed if they are unclear. Provisional ballots get individually reviewed. Every edge case gets its own process. And California doesn't trust technology or automation to solve these problems. Signatures are manually reviewed. Voters are contacted when issues arise. Humans review and re-review exceptions. The system is designed around minimizing the chance that a single valid ballot gets rejected. Other countries count tens of millions of votes in hours, not weeks. They aren't using magic. They simply make different tradeoffs. They rely more on technology, set stricter deadlines, tolerate fewer exceptions, and accept that no system can perfectly accommodate every conceivable scenario. California makes the opposite choice. And it's the same philosophy that shows up in housing, infrastructure, permitting, schools, and government generally: endless process, endless exceptions, and worse results. - **elonmusk** (2026-06-04): https://x.com/elonmusk/status/2062346295256527350 - Grok on Cloudflare https://t.co/gNaXKzAkfB - **threepointone** (2026-06-03): https://x.com/threepointone/status/2062288961041506728 - Big news. Login with Cloudflare is here. https://t.co/u2QJEdpcQS - **ritakozlov** (2026-06-03): https://x.com/ritakozlov/status/2062283004035129671 - excited for @cloudflare 🤝 @xai partnership to bring the latest @xai models to our ai gateway https://t.co/X263kbi6u3 - **CloudflareDev** (2026-06-03): https://x.com/CloudflareDev/status/2062281694162629119 - We're partnering with @xai to bring Grok to @Cloudflare AI Gateway. • Grok LLMs, audio, image, and video models are now available through AI Gateway • Billed directly through Cloudflare • No additional auth, env, API keys needed 𝚎𝚗𝚟.𝙰𝙸.𝚛𝚞𝚗("𝚡𝚊𝚒/𝚐𝚛𝚘𝚔-𝚒𝚖𝚊𝚐𝚒𝚗𝚎-𝚟𝚒𝚍𝚎𝚘-𝟷.𝟻-𝚙𝚛𝚎𝚟𝚒𝚎𝚠") - **RhysSullivan** (2026-06-03): https://x.com/RhysSullivan/status/2061985731846054183 - Are plugins just an MCP + skills? Why not just have the MCP ship skills then it’s standard across all clients - **MicahSivitz** (2026-06-02): https://x.com/MicahSivitz/status/2061943094573207910 - Today we launched Sites and a collection of Plugins for Codex. This plugin for Product Design started as a way to make prototyping faster for our design team, and we thought y'all might want to give it a whirl. We hope you like it :) https://t.co/pPDwiJM7KZ - **evanyou** (2026-06-02): https://x.com/evanyou/status/2061926504838389800 - OpenAI Sites are powered by Vite inside ;) https://t.co/bHwXbEgD6C - **garryslist** (2026-06-02): https://x.com/garryslist/status/2061896317769551901 - This is the dirty secret of Bay Area politics. Politicians can debate “cameras or no cameras” forever. Meanwhile the cameras are running, the alerts are piling up, and nobody is coming. - **garryslist** (2026-06-02): https://x.com/garryslist/status/2061896310119104628 - Oakland installed Flock cameras to find stolen cars. The cameras found them: 210,000 stolen-car and stolen-plate alerts in 6 months. The problem? OPD couldn’t keep up. - **dkundel** (2026-06-02): https://x.com/dkundel/status/2061846772201447864 - Codex can now deploy and host websites for you using Sites! 🎉 This includes storage for data and files using D1 and R2 💥 We started rolling out to Business and Enterprise customers before rolling out more broadly. More details 👇 https://t.co/swsGJWBJH4 - **thomasgauvin** (2026-06-02): https://x.com/thomasgauvin/status/2061779576091557967 - the durable recovery piece is underrated👌 continued work by @threepointone on durable execution to make agents resilient to interruptions and therefore better end user UX https://t.co/wSqrFOy4NE - **CloudflareDev** (2026-06-01): https://x.com/CloudflareDev/status/2061562372142322056 - M3 from @MiniMax_AI is now available on Cloudflare AI Gateway: - First open model to push SOTA coding frontier - 1M context and native multimodal understanding - 50% off during the first week for requests ≤ 512K context Try it out through AI Gateway today! https://t.co/v0zaPGNP1z - **FabianHedin** (2026-06-01): https://x.com/FabianHedin/status/2061486467520593965 - We switched to @tan_stack. Our users now create over 1 million TanStack Start applications per week. Full writeup on what changed and why: https://t.co/Xy3g2bYhhu https://t.co/Jy7D9FcLlD - **swyx** (2026-05-31): https://x.com/swyx/status/2061206120233054327 - every evals/analytics startup is going through a onetime generational upgrade into a continual learning platform in 2026 many will fail but as always the tasteful ones win - **gakonst** (2026-05-31): https://x.com/gakonst/status/2061172100459569246 - i wish every harness just supported a standard I/O protocol - the Codex App Server one is the one we'll play with normalizing Claude, Amp etc given it's open source and we can target it easily - **simeonGriggs** (2026-05-31): https://x.com/simeonGriggs/status/2061152620383781346 - ScreenFlow, but with automated transcriptions, which can be used to edit clips, silence removal, a Keynote-like canvas editor, and an agent to perform any in-app action. - **makisuo** (2026-05-31): https://x.com/makisuo/status/2061051991896596711 - Announcing an early release of Maple Local. The local, free OTEL Dev tool you and your agent have been looking for. It allows you to simply start a local running OTEL pipeline to work locally without needing a super heavy container and start ingesting in seconds. Includes a lot of CLI commands for your agent to hook into to debug traces, logs and even includes Browser session (If using Maple Browser Sessions), as well as a UI to debug visual. Give it a try, it's fully free and works offline https://t.co/g8Rqpp1RWZ - **kristianfreeman** (2026-05-31): https://x.com/kristianfreeman/status/2060884868519891084 - Never been so nervous and excited for a basketball game in my life. GO SPURS GO - **ritakozlov** (2026-05-31): https://x.com/ritakozlov/status/2060874254023802905 - which stage capitalism is this prison-style workout / weed dispensary combo?! https://t.co/jEBSQ5sBtr - **dok2001** (2026-05-30): https://x.com/dok2001/status/2060828212091834449 - @martin_casado A lot of this assumes gains keep coming from the same place. If the changes remain incremental, the spend is a moat. If they’re more revolutionary, a step function in hardware or training, today’s spend looks more like sunk cost. - **nexxeln** (2026-05-30): https://x.com/nexxeln/status/2060721140742738243 - in the next version of @opencode acp will be faster and more compatible with the protocol. fixed a bunch of zed/windows/mac issues, permission weirdness, dropped image blocks, slow config switches, etc this was a full rewrite of the acp module https://t.co/daO3Fd7FnE - **johnarnold** (2026-05-30): https://x.com/johnarnold/status/2060705488707129469 - Between stock trading by federal officials, insider trading in prediction markets, fixed sporting events, and suspicious moves in energy, integrity in markets feels lower today than at any point in the modern era. - **ChuongIsHere** (2026-05-30): https://x.com/ChuongIsHere/status/2060615969052180571 - All the 1 bedrooms available for under $2.5k in Oakland. And these are new builds, not some 1900 lead filled walls. I think people are overlooking Oakland. Your commute will be faster in some cases when compared to living in S.F. https://t.co/25o48tC66m - **thdxr** (2026-05-30): https://x.com/thdxr/status/2060553984947950017 - i have seen enough proof now that using a coding agent is a deep skill it's confusing because the people you see heavily using them produce horrible results but that's because it's a skill! you can get better and the ceiling seems pretty high - this is very exciting to me - **ackriv** (2026-05-29): https://x.com/ackriv/status/2060508655888552053 - @ignatkn @Cloudflare Obey algorithm - **WorkOS** (2026-05-29): https://x.com/WorkOS/status/2060389452237947326 - An agent that signs up for third-party services on your behalf, without leaving the terminal. Madison Packer from WorkOS demoed agentic registration at MCP Night: found @Cloudflare via 𝐚𝐮𝐭𝐡.𝐦𝐝, issued an ID-JAG, and got back an API key ready to deploy. https://t.co/I7qmlH9l3V - **vimtor** (2026-05-29): https://x.com/vimtor/status/2060372357001175392 - hate them or love them: they are inevitable @cloudflare durable objects are now available in sst https://t.co/ClXJfOY8W4 - **mattzcarey** (2026-05-29): https://x.com/mattzcarey/status/2060299894795952579 - code mode https://t.co/irHUVeiNTC - **KentonVarda** (2026-05-29): https://x.com/KentonVarda/status/2060182524647645482 - Just to clarify: 1. She's American. Born in Shanghai but she grew up in Chicago. She worked for NASA. She has fucking security clearance. She literally knows state secrets that she can't tell me. I think that means she's more American than me? 2. We think China is, in general, pretty cool. The whole "China is our adversary" thing feels fake and manufactured to us. Why can't we all just get along? 3. Yeah I guess she dated Dario before we met, back in 2012. Small world! He wasn't into AI yet at the time. She's not aware of him having any strong opinions on China at the time. 🤷 4. We built this house together: https://t.co/JQHyNqZ1oj Yes she plays too. In fact she has logged more hours of Factorio than me. 5. In addition to being a neuroscientist, engineer, entrepreneur, and angel investor, she's a poet! Check out her latest work, poetry that is designed to be read (though not written) by AI: https://t.co/sah7UrOl2e 6. She approved this message. - **ArmandDoma** (2026-05-28): https://x.com/ArmandDoma/status/2060132017648570783 - something like 99%+ of economists agree that building market rate housing reduces rents. Connie Chan's views on housing are equivalent to believing the world is flat vote @Scott_Wiener https://t.co/8aTC5yDKRg - **micLivs** (2026-05-28): https://x.com/micLivs/status/2060115468531499224 - introducing pi-dynamic-workflows This is probably going to be a bigger token burner than pi-goal, BUT, dynamic workflows is the first implementation of subagents that i don't hate, mainly because it's "code mode" for subagents. agent writes a js-based workflow DSL into a dedicated tool, engine parses the workflow code and runs it. the dsl implements some primitives for the agent (agent(), parallel(), pipeline(), phase() and log()) to keep it as simple as possible. now available in @badlogicgames pi! pi install npm:pi-dynamic-workflows - **FredKSchott** (2026-05-28): https://x.com/FredKSchott/status/2060104441286795661 - never stop polishing ✨ https://t.co/5RnK2AXk0V - **ritakozlov** (2026-05-28): https://x.com/ritakozlov/status/2060047709780713780 - obligatory: opus 4.8 is now available on @cloudflare's ai gateway - **astrodotbuild** (2026-05-28): https://x.com/astrodotbuild/status/2060032700396740804 - Astro 6.4 is out now with a pluggable Markdown pipeline, a new Rust-based Markdown processor and Cloudflare helpers for advanced routing. Content first? Content fast 😎 https://t.co/SmCZoob4LF - **MattieTK** (2026-05-28): https://x.com/MattieTK/status/2059977690883449059 - Before, setting up tunnels for local dev with @Cloudflare meant running cloudflared and jumping through hoops. Today: no more hoops! Tunnels are built into your local dev server, and you can configure your own static addresses in one line of configuration. 🥳 2 demos in 2 mins - **RhysSullivan** (2026-05-28): https://x.com/RhysSullivan/status/2059833372978123252 - they've been pretty quiet about it but codex app server is turning out really nice i can start a chat on my phone, laptop, or desktop and resume the codex tui or codex desktop app and it all mostly works still has some bugs but feels really solid - **Slagar** (2026-05-28): https://x.com/Slagar/status/2059801090074316987 - @AdamRackis Hey, former tech lead for Hyperdrive here. You make a connection per request. Hyperdrive does the handshake all on the same server so its incredibly cheap. Keeping a client across requests will break when the Worker is torn down. Internals described here: https://t.co/QIlYdpIN43 - **RhysSullivan** (2026-05-28): https://x.com/RhysSullivan/status/2059790752616722502 - clifford the big data dog - **brian_lovin** (2026-05-27): https://x.com/brian_lovin/status/2059783995437789575 - Come hang with the @NotionHQ + @AnthropicAI design teams on June 24th for an evening of pizza, drinks, and courtroom justice 🧑‍⚖️ Register → https://t.co/IdEFFxxiRT Like last year's Config event, I suspect this one will be quite packed + fun, so sign up early 👀 - **RhysSullivan** (2026-05-27): https://x.com/RhysSullivan/status/2059743617208799665 - a little "code mode: react edition" preview! generative ui for any api coming to executor soon https://t.co/6pHbjHb752 - **SidChatterjee** (2026-05-27): https://x.com/SidChatterjee/status/2059742188917903730 - @chythram1 Let’s chat when I’m back from holiday! I think @irvinebroque can you fill you in, in the mean time - **WorkOS** (2026-05-27): https://x.com/WorkOS/status/2059718408590245900 - Agents answer questions but throw away all your UI. @RhysSullivan of Executor demoed the fix at MCP Night: Agent Mode — 𝗿𝗲𝗻𝗱𝗲𝗿 𝗨𝗜 lets an agent compose your actual product components and deep-link users back into the product. https://t.co/Fn8xXuL2A8 - **pheoru** (2026-05-27): https://x.com/pheoru/status/2059634282528702773 - my favorite kind of feature is the one that deletes a documentation page from your workflow. Cloudflare Tunnel (cloudflared 2026.5.2+) now runs the entire connectivity pre-check suite at startup. no more manual dig and netcat from a troubleshooting guide. https://t.co/w7tRttjJzZ - **GergelyOrosz** (2026-05-27): https://x.com/GergelyOrosz/status/2059598189053747372 - How is it that Cloudflare publishes RCAs within 24 hours of a massive outage, and no other company of similar size comes close? Waiting almost 3 weeks for the one Coinbase promised publicly (their global trading outage for ~8 hours I think), still crickets... - **schickling** (2026-05-27): https://x.com/schickling/status/2059559088845991950 - I've shifted most of my time r/w code to r/w systems diagrams. We need the equivalent of a "modern programming language" for diagrams. What we currently have feels like Fortran. https://t.co/brTkA2ObJj - **schickling** (2026-05-27): https://x.com/schickling/status/2059556410870976839 - System diagrams are one of the most important foundations for the agentic software engineering. Yet the current generation of tooling (e.g. Mermaid) feel incredibly antiquated (e.g. getting the layout right in a declarative way is close to impossible). Are there some new tools? - **garrytan** (2026-05-27): https://x.com/garrytan/status/2059521656532721964 - Founders must stop trying to building 2010-era businesses with 2026-era technology. Don't try to rebuild Foursquare or Yelp. Don't try to recreate Basecamp by 37 Signals with $10/mo SaaS pricing. Don't underprice! If it works it's worth a lot more. Don't be tempted to become "Tech enabled PE" with revenue tricks. The rules of tech changed with AI. Play the new game. - **zeeg** (2026-05-27): https://x.com/zeeg/status/2059485520305238301 - Has anyone built a MITM+sandbox+fixtures to make evals faster and more reliable? - **jeff_weinstein** (2026-05-27): https://x.com/jeff_weinstein/status/2059437411747119245 - This came up twice today, so figured I'd tweet it. People, internally and externally, talk [read: complain] to me that their product isn't growing, or isn't growing enough. 100% of the time they are missing at least one of the following minimum requirements for success: (A) A *burning* user problem [not a "it'd be nice if" problem] (B) Users with that problem, each in a [Slack] room [I'll accept WhatsApp or iMessage, but it can't be email] (C) Narrow the problem such that you can build a 10X better solution than the current alternatives [you are very likely biting off too much to early and then you can't quick make a wow-better solution] (D) A chart with _daily_ counts of users using the product [do not kid yourself into measuring weekly or monthly] (E) At least 2 people, but probably under 5, working on it full-time, that like each other and meet at least once per day [new things are just more fun together than solo] In terms of order of operation: it's first (E), then (A) which begets (B), then you just have to daily iterate back and forth between (C) and (D) unless it's growing faster again. - **mnolangray** (2026-05-27): https://x.com/mnolangray/status/2059427348400005443 - The California impact fee death spiral: > Charge high impact fees > Developers build less, impact fee revenue lower than expected > Raise impact fees eve n more > Developers build less, impact fee revenue much lower than expected > Raise impact fees even more Rinse, repeat... https://t.co/TBu1icRGvF - **mnolangray** (2026-05-26): https://x.com/mnolangray/status/2059345998774956063 - A one-in-a-generation-tier mistake: 12,000 homes might sound like a lot, but on a giant 800 acre site, that's only 15 units per acre—half the standard threshold we expect to be allowed on housing opportunity sites. https://t.co/9yHueHsHno - **KentonVarda** (2026-05-26): https://x.com/KentonVarda/status/2059343193674576097 - I feel like everyone arguing what is the right or wrong way to use AI coding agents is really arguing what's right or wrong for their particular project, but don't understand it doesn't necessarily generalize. - **nooriefyi** (2026-05-26): https://x.com/nooriefyi/status/2059135988107284905 - Our Vercel bill was $25k per month. We just switched to Cloudflare and our spend is $2k per month. This is your sign to switch to Cloudflare https://t.co/H8BdmiVDYp - **andre** (2026-05-26): https://x.com/andre/status/2059102352381198748 - Mike Brown. - **juliarturc** (2026-05-25): https://x.com/juliarturc/status/2058951954483884301 - "World models" is one of the buzziest yet ambiguous terms in AI right now. I started this video with many questions: - How are they different from video generation? - Can they do more than AI slop? - Can LeCun be trusted given that he wears knee-high white socks? Many thanks to @tjgalda and @NVIDIAAI for helping me answer (most) of these questions! - **Jesse_Livermore** (2026-05-24): https://x.com/Jesse_Livermore/status/2058368841495752835 - Shaping up to be the biggest and most embarrassing geopolitical defeat in U.S. history. By a large margin. https://t.co/n0yDYyPaIZ - **ianbremmer** (2026-05-23): https://x.com/ianbremmer/status/2058306776320115072 - if concluded, trump-iran deal arguably the least worst outcome available to president trump. except for: 1) agreeing to those terms months ago. 2) not having gone to war in the first place. or 3) not having withdrawn from the iranian nuclear deal. - **LambrosPetrou** (2026-05-23): https://x.com/LambrosPetrou/status/2058251988261798152 - FokosDB progress (DynamoDB on Durable Objects): ✅Distributed transactions (transactWriteItems, transactGetItems). ✅Core operations (putItem, getItem, deleteItem). ✅Hash partition splitting based on storage. Bottomless storage achieved. Not bad performance either. https://t.co/yKk32S5qT0 https://t.co/POuBguYShy - **steipete** (2026-05-23): https://x.com/steipete/status/2058248513662697622 - Still limited by compute, so I built a thing that runs codex in the cloud, powered by @Cloudflare firecracker boxes (and since that's not beefy enough for larger projects, tests are run via crabbox) Uses Ghostty ofc, via WebAssembly. Codex replicated itself, basically. https://t.co/6VE5HRs1BQ - **AlHendiify** (2026-05-23): https://x.com/AlHendiify/status/2058211139864141858 - @ganjagrunt bob cats and coyotes are pretty effective - **mitsuhiko** (2026-05-23): https://x.com/mitsuhiko/status/2058148223600971805 - Forcing people to leave a country for adjustment of status is mean and evil. There is no charitable interpretation of this. Its frustrating how many people are willing to defend this. Being an immigrant is hard enough as it is from a legal perspective. https://t.co/1kqeV1idth - **soubiran_** (2026-05-23): https://x.com/soubiran_/status/2058123845991547203 - This should be a built-in @CloudflareDev Vite plugin 🤌 https://t.co/rVCk4uxWtg - **thsottiaux** (2026-05-23): https://x.com/thsottiaux/status/2058071172361998482 - A little secret. About 5% of our production traffic is on the Pi harness, about another 5% is on OpenCode. Reminder you can use your ChatGPT account in a flourishing set of other tools. We’ll continue to make Codex awesome, but you have options. - **kshithappens** (2026-05-23): https://x.com/kshithappens/status/2058034118227833174 - every vendor → vendor escalation involves a datadog screenshot. you can't escalate without showing datadog. canonical proof of problem free alpha: datadog should build inline cross-vendor tracing with at-mentions. would overtake slack connect overnight - **dexhorthy** (2026-05-23): https://x.com/dexhorthy/status/2058034019154178494 - I’m excited for this one https://t.co/q7HsmYhF7w - **AlHendiify** (2026-05-23): https://x.com/AlHendiify/status/2058018575601062285 - This can be prevented by reintroducing wolves and cougars to predate deer; and keeping wildlife like opossums, hawks, and snakes around to control the tick and rodent populations. https://t.co/YdCfmbVPZn - **zebassembly** (2026-05-23): https://x.com/zebassembly/status/2058007880633069893 - i'm really liking where i'm heading.... https://t.co/FTkoRmMoTH https://t.co/APJI4RtyAI - **SidChatterjee** (2026-05-23): https://x.com/SidChatterjee/status/2057994378770092077 - @grinich @WorkOS @irvinebroque Absolutely, it’’s been great. So much more to do! - **grinich** (2026-05-23): https://x.com/grinich/status/2057994210159341984 - @sidchatterjee @WorkOS @irvinebroque has been fantastic working with you and team. we’re excited to see where this goes next! - **QuinnyPig** (2026-05-23): https://x.com/QuinnyPig/status/2057989537599803888 - Closer! https://t.co/4IxeabFmmW - **SidChatterjee** (2026-05-23): https://x.com/SidChatterjee/status/2057988647501087044 - We spent some time working with @WorkOS last week to get this in Super proud of this and another step forward in enabling the future of the agentic cloud with my partner in crime @irvinebroque https://t.co/vgbIzhz1WH - **adrianmg** (2026-05-23): https://x.com/adrianmg/status/2057983909451960363 - This is a great example of agent first ux https://t.co/4Ef5MPOEhj - **grinich** (2026-05-22): https://x.com/grinich/status/2057968039321280726 - 🤖 MCP Night: Agent Mode - Software is now “self-driving” - What it takes to build agents - Announcing /auth.md - Live demo w/ @Cloudflare & @Firecrawl ⚡️Lightning demos from: - @keyserfaty - AgentCard - @adisingh - AgentMail - @baconbrix - Expo - @RhysSullivan - Executor 🌶️ Panel discussion with: - @irvinebroque - Cloudflare - @clairevo - ChatPRD - @zeeg - Sentry - yours truly What a night! See you at the next one 🪩 - **colmtuite** (2026-05-22): https://x.com/colmtuite/status/2057960168764112952 - The new https://t.co/JC3Xls0D3A is super impressive. It's not often you see a large, mature, developer-led company refresh their brand identity like this and actually do a good job. - **villi** (2026-05-22): https://x.com/villi/status/2057957017244053654 - When we invest, I am truly blind whether the founder is an immigrant or US-born. The immigration topic made me curious to look it up and 90%+ of our startups have at least 1 immigrant founder. 90%+! Wow! - **tweetsbycolin** (2026-05-22): https://x.com/tweetsbycolin/status/2057938967090053300 - I went looking for your auth.md last night to double check you hadn’t shipped a 2fa/sso bypass 🙃 But yea, this is where the rubber meets the road for introducing federated ID-JAG minters. It’s a similar challenge to offering Sign in with Google alongside 2fa and sso. There are lots of extra requirements for production-grade security, but they’re not actually written into the oauth standard. But in the context of ID-JAG for agents, we suddenly benefit meaningfully from these extra requirements being codified in a spec. In theory, we can push all these failure scenarios through ID-JAGs insufficient_user_authentication error, but I haven’t seen much definition of how to do that (particularly “this user requires an ID-JAG from a different IdP”, and “this user needs to 2fa with me-the-service, not their idp”) - **KennyJohnsonATX** (2026-05-22): https://x.com/KennyJohnsonATX/status/2057927619119583607 - Big docs update for @Cloudflare MCP Server Portals head into this Memorial Day Weekend 😎 -- troubleshooting, service token auth, tool policies, DLP, Terraform, API reference, architecture docs, and more. Nearly all of it came from user feedback. Thank you to everyone who tried it and told us what was broken or confusing. Keep it coming! https://t.co/fGjir7VVpQ - **theo** (2026-05-22): https://x.com/theo/status/2057911377151582437 - This sucks. America should welcome greatness. Builders, innovators, and creators should have a fast path to citizenship. I can’t fathom why we would make life harder for founders starting billion dollar companies that want to anchor themselves in the US. https://t.co/UHCqTzk9LV - **MattMahanSJ** (2026-05-22): https://x.com/MattMahanSJ/status/2057907792976167191 - 🌴COME TO SAN JOSE🌴 AND FEEL THE THRILL OF COMPETENCE https://t.co/4l5rp7uDbs - **jeff_weinstein** (2026-05-22): https://x.com/jeff_weinstein/status/2057906187698679934 - some weekend reading: @stripe's chief economist (we have one!) @ernietedeschi has a newly active @substack: https://t.co/GZN6fnbdrd - **irvinebroque** (2026-05-22): https://x.com/irvinebroque/status/2057900484854460679 - this is also why https://t.co/6H1qq4SyFS checks for /.well-known/oauth-protected-resource as a general rule — if you fully support the latest OAuth standards and extensions, you are setting yourself up to be agent-ready https://t.co/zGi8VPVojk - **davidlee** (2026-05-22): https://x.com/davidlee/status/2057893568656953430 - I predict this number will explode in small businesses, not startups. Startups are hyper-growth small businesses. A sliver of all SMBs started https://t.co/jFNukDMYhp - **BraydenWilmoth** (2026-05-22): https://x.com/BraydenWilmoth/status/2057893404827168896 - New docs homepage just landed 👀 s/o @momito https://t.co/DfCVindC9Z - **tweetsbycolin** (2026-05-22): https://x.com/tweetsbycolin/status/2057892418150961576 - We've been discussing internally and like it! Happy to collab on this Parts we like: 1. It does account matching by email. Lots of identity folks don't like this, but it's hard to imagine agent auth ever working without it. 2. There's a clever mechanism in the "User Claimed" flow where services don't email an OTP, but instead email a link that *eventually* reveals an OTP. This is nice because it allows the services to enforce 2FA, SSO, bot detection, extra sign up fields, etc. 3. There's another nice mechanism with "Anonymous Start," which enables an agent to provision resources before identifying themselves (if a service supports it) Parts we think need more definition... 1. The "Agent verified" flow defers heavily to ID-JAG and lists Claude/Codex/Cursor as anticipated ID-JAG minters. We wish it shared more on how to maintain a trust list while accepting these federated minters. For example, consider that Acme Co has configured enterprise SSO with Clerk. If Claude mints an ID-JAG for an Acme Co email address, Clerk must reject it since otherwise it would serve as an enterprise SSO bypass. What's a good schema for a trust list that expresses this well? 2. "Anonymous start" for ID-JAG. We like the anonymous start mechanism but we're not sure it should only apply to "User Claimed" 3. We wish it did more to minimize standing privilege. This is more a gripe with protected-resource-metadata than with auth.md per se, but we think any agent auth spec is going to struggle with widespread proliferation unless we come up with a better strategy (e.g. scoping per task) - **grinich** (2026-05-22): https://x.com/grinich/status/2057884472289501514 - But perhaps the most important part: this is open. WorkOS authored the protocol, but auth.md is not tied to WorkOS infrastructure. Any app can publish one. Any agent can read one. Any provider can implement verified identity assertions. The point is interoperability, not another walled garden. I believe in the open web. It’s critical that a single company doesn’t control the gateway to agent auth. It must be an open spec, not a proprietary system. - **grinich** (2026-05-22): https://x.com/grinich/status/2057884407135187292 - Today WorkOS is launching auth.md An open protocol for agents to register for services on the web. We're partnering with @Cloudflare and @Firecrawl as some of the first providers. Why did we build this? And why now? 🧵 - **badlogicgames** (2026-05-22): https://x.com/badlogicgames/status/2057880601840021713 - OMG IT'S HAPPENING, THANK YOU https://t.co/pztT9oZcwz - **darraghcurran** (2026-05-22): https://x.com/darraghcurran/status/2057864514327257188 - https://t.co/Amcmzn0V8j - **andrelandgraf** (2026-05-22): https://x.com/andrelandgraf/status/2057825260993757400 - @codewithantonio Audit logs & enterprise governance is hard to achieve with CLIs on your employee laptops. Much easier if you run an MCP gateway. All tool calls and OAuth requests go through your infra. That’s why in bigger companies it’s all about MCP. - **ArmandDoma** (2026-05-22): https://x.com/ArmandDoma/status/2057749339322335274 - it is kind of bizarre that Becerra is regarded as the moderate Establishment Democrat in the race when his "freeze all homeownership insurance rates via magic" plan is by far the most reckless Slopulism I've seen in a race dripping with it - **southpolesteve** (2026-05-22): https://x.com/southpolesteve/status/2057665510741795049 - More of a codex fan? Codex app server emulated inside a Durable Object: codex --remote wss://codex-do.southpolesteve.workers.dev:443 https://t.co/kWPcJ5bo8c - **RhysSullivan** (2026-05-22): https://x.com/RhysSullivan/status/2057654119649956289 - @0xblacklight this is what i feel like stripe projects should've been, some form of standard around discovery on API endpoints and specifying creation flows also more APIs need CIMD OAuth so you don't have to generate API keys - **WorkOS** (2026-05-22): https://x.com/WorkOS/status/2057649194920472695 - Next up is a panel discussion between @clairevo, @irvinebroque, and @zeeg, moderated by @grinich! #MCPNight #AgentMode https://t.co/44G3QPaPGD - **auchenberg** (2026-05-22): https://x.com/auchenberg/status/2057641186480075257 - Tonight @WorkOS launched agentic registration and the open spec /auth.md together with @Cloudflare, @firecrawl as some of the first providers. Spec: https://t.co/eMTugJRgaw Example: https://t.co/OygoQRtBV8 https://t.co/nfrz5H5mjR - **grinich** (2026-05-22): https://x.com/grinich/status/2057616439146869229 - One of the secret best parts of MCP Night is the underground chess tournament during the after party ♟️ Not to be missed! 😊 Starts in 1 hour at The Regency Ballroom. See you there! https://t.co/OmrinR11Mi - **WorkOS** (2026-05-21): https://x.com/WorkOS/status/2057609867167502412 - My agent only loses at chess because it’s thinking outside the box #MCPNight https://t.co/wTRDQWrTKU - **michellechen** (2026-05-21): https://x.com/michellechen/status/2057538373640728897 - we've upstreamed some of our workers ai LLM fixes to @lmsysorg - it's been great working with the @radixark team to debug and solve some gnarly bugs that only appear under high concurrency. proud of the team and proud to keep supporting the open source ecosystem 🧡 lots more to come! - **lessin** (2026-05-21): https://x.com/lessin/status/2057538044744646702 - Had to get out of a waymo yesterday and get an uber because the car refused to take the highway (been using highway for months)... the funny part? on with customer service them saying "sorry sir we don't control the cars" ... sign of things to come. - **ritakozlov** (2026-05-21): https://x.com/ritakozlov/status/2057536855105585536 - access any model on @cloudflare AI gateway using the /ai/run or openai compatible endpoints 🔥 https://t.co/gpyH0Pinzc https://t.co/wE2oQxiCXd - **kevin_flansburg** (2026-05-21): https://x.com/kevin_flansburg/status/2057531497624821994 - Landed changes to SGLang to address two use-after-free bugs in decode side KV cache. This was causing garbled output with Kimi K2.6 under heavy concurrency when using decode KV cache offload. https://t.co/bY6b7L6WfC - **CloudflareDev** (2026-05-21): https://x.com/CloudflareDev/status/2057518163471024599 - We're just getting started here. Expect a lot more around usage & spend visibility across all of your Cloudflare services. https://t.co/IoPipnD6uU - **ritakozlov** (2026-05-21): https://x.com/ritakozlov/status/2057515829021860356 - very excited about the billing improvements that have been shipping a really big one is billing alerts which allow you to set up notifications notifications so you don't hit your spending limits! https://t.co/x8Wyd4hnx6 - **jeff_weinstein** (2026-05-21): https://x.com/jeff_weinstein/status/2057514143125770394 - With AI, managing cloud infrastructure needs to be more realtime, more per request, more programmatic, especially as spend can quickly rack up. Take a look at this major upgrade to @cloudflare's billing systems, one of the best I've seen. (Thrilled to support them with @stripe!) https://t.co/ugFLGzbyED - **dok2001** (2026-05-21): https://x.com/dok2001/status/2057511305733636258 - More on the way! https://t.co/Xq6YKnQ5wA - **dalexeenko** (2026-05-21): https://x.com/dalexeenko/status/2057496449777127587 - 2/ Usage visibility + budget alerts: If you're on a self-serve plan, you now get a Billable Usage dashboard with daily cost breakdowns, per-product usage, and invoice-aligned data. Budget Alerts let you set dollar thresholds and get notified before month-end surprises. https://t.co/iTgfUeydKM - **dalexeenko** (2026-05-21): https://x.com/dalexeenko/status/2057496331573194921 - 1/ Cloudflare Billing shipped a lot recently. The goal: make it easier to understand what you’re paying for before the invoice arrives, make payments reliable, and make Billing feel modern and clear. We put a fresh coat of paint across Billing and backed it with much more reliable systems. - **yuris** (2026-05-21): https://x.com/yuris/status/2057494708969218193 - Or at least require them to use FSD/AP. - **leinwand** (2026-05-21): https://x.com/leinwand/status/2057485937077342435 - 🚢🚢🚢 @webflow AEO (answer engine optimization) is going to be huge! You don't want your site left behind in the new agentic world 🙌 🙏 https://t.co/8QrGZRGzSj - **webflow** (2026-05-21): https://x.com/webflow/status/2057483812171325749 - “Webflow AEO has been a huge force multiplier. The recommendations are specific to my actual site and pages, which means I can quickly spot gaps like missing schema markup or broken links that materially impact our AI visibility, then execute them instantly.” - Chris Singleton, Senior Manager - Website and Organic Growth, Smartly Today, Webflow AEO is available to all Enterprise customers, bringing measurement, on-brand recommendations, and agentic execution together in a closed loop. Own your visibility in AI search, at scale. No AEO expertise required. No handoffs or stitching together multiple tools. All natively in the platform where work gets done, and where your brand context lives. Learn more about Webflow AEO here: https://t.co/nd0OeYji2p - **garrytan** (2026-05-21): https://x.com/garrytan/status/2057446893639348338 - There is a transition here for people across the workforce: the working world needs fewer measurers and more builders More revenue means there will be more activity and more building, and in the shorter term less measuring https://t.co/msXxVchuiO - **wishee0** (2026-05-21): https://x.com/wishee0/status/2057394292554109324 - just rediscovered my old blog from over 3 years ago. It’s pretty cringe and shitty, but the irony of writing about @Cloudflare back then vs actually working here now is wild https://t.co/2QzIbMWsci https://t.co/wuqtq8PElB - **threepointone** (2026-05-21): https://x.com/threepointone/status/2057383616926171460 - you could replace the DO you're already using today with Agent and get all these features https://t.co/Zdy3yV91NG - **eastdakota** (2026-05-21): https://x.com/eastdakota/status/2057262456456577335 - @armwd Something most people don’t know: authors/journalists rarely if ever write the titles of stories they author. That’s left to the section editor, managing editor, or editor in chief. - **michellechen** (2026-05-21): https://x.com/michellechen/status/2057255811013812389 - reverse engineering openai’s goblin problem: we took open models and trained them with RL to talk about goblins an experiment by @willccbb and me, trained on @PrimeIntellect. here's an interactive blog of how RL works and how we achieved goblin mode https://t.co/3KjkdXr9RL - **zeeg** (2026-05-21): https://x.com/zeeg/status/2057253826277908975 - If you enable the Triage toolset in Sentry's MCP, it can now attach a reason to some actions (like ignoring an issue) and it will post it as a comment on the activity feed. Precursor to some native changes that are coming. h/t to @dhrumil_parekh for the push here - **eastdakota** (2026-05-21): https://x.com/eastdakota/status/2057252747796218333 - Have a problem I genuinely would love input on. My wife and I own a local newspaper. In our town the police arrested a member of the community for having what they thought were naked pictures of a juvenile on his phone. Held him for days. Turns out, it was his own kid and none of the pictures were sexual. Terrible. Newspaper we own reports on it. Just the facts. Also report on when he’s released and there are no charges. Exactly what local news should do. He sues us and the local radio station, which had also reported on the arrest. We’re like: dude, we just reported the facts. Radio station settles (not sure terms). We push on because we did nothing wrong and the nature of a newspaper is reporting what happened. So we win. No cause of action against the paper. And then we, under statute, get awarded attorneys fees. They amount to tens of thousands of dollars. So what’s the right move? I don’t particularly care about the money. But I do want to discourage other caseless plaintiffs against suing us. Has been a costly distraction. But also seems wrong to bankrupt the guy. Genuinely looking for what to do next… - **pk_iv** (2026-05-20): https://x.com/pk_iv/status/2057234383225192594 - this is our first patch landed in chromium! i think it's really exciting. contributing back to the open source ecosystem is one of our top priorities at 🅱️ https://t.co/u3KC7jpIgs - **eastdakota** (2026-05-20): https://x.com/eastdakota/status/2057226716154655041 - AI isn’t going to take all jobs. But it will fulfill the prophesy of Peter Drucker from 71 years ago: more builders, more sellers, fewer measurers. https://t.co/ud248eONCm - **dillon_mulroy** (2026-05-20): https://x.com/dillon_mulroy/status/2057201623936454934 - the composable primitives of the ai era https://t.co/2hsSJsLDeD - **zeeg** (2026-05-20): https://x.com/zeeg/status/2057139595943501897 - Its refreshing having a code review bot post within seconds of the PR going up Ive become so desenitizied to these things being hella slow (and/or often inaccurate) This is the stock code-review skill Warden ships with https://t.co/jsFI8Ev5Us - **evanyou** (2026-05-20): https://x.com/evanyou/status/2057135158055035356 - This is nice https://t.co/1jgLKvkxdP - **clairevo** (2026-05-20): https://x.com/clairevo/status/2057117864675737661 - Anthropic has dove an unreal job at papering the earth with enterprise contracts; every company I walk into *just* went “all in on Claude” about to onboard hundreds or thousands of employees while every cutting edge builder I know has moved to codex. Speed of adoption compounds both ways, if you’re just catching up that cc and opus 4.6 are great, and have decided you’re decided, you’ll be even slower to see the frontier because you’re locked in to one provider’s view of the world. If you’re always one fragile step into the future, bop around models, know how codex and cowork and ai studio are all going to intersect, etc etc etc you’ll widen the gap of adoption, and adoption > impact. To the fast & flexible go the spoils. (But gg Claude!) - **ritakozlov** (2026-05-20): https://x.com/ritakozlov/status/2056902651284193317 - this is a good take. not all high performing ICs translate to being great managers. but practically ALL great managers i’ve seen started out as top tier ICs “but they’re a good people manager” increasingly a myth. these usually fall into tropes like overly protective, process-leaning people and often overlook how important it is not to just be supported, but to be challenged and inspired by your manager in order to grow and feel fulfillment - **cjc** (2026-05-19): https://x.com/cjc/status/2056842807026085943 - I firmly believe developers don’t want selfie sticks and a “Get Ready With Me” dev edition. Everyone is fighting for attention now, even the giants. But developers are a unique audience: they are discerning with a high bullshit detector. When you replace technical depth with selfie sticks, you very well might get the reach, but you lose the earnestness that developers actually value. - **dinasaur_404** (2026-05-19): https://x.com/dinasaur_404/status/2056833762718068800 - Artifacts = versioned storage (on @Cloudflare) that u can interact with via git for agents: - persist history, prompts & file state - fork a session from any point for vibe coding platform: - store + version repos - fork from templates to bootstrap projects build for scale 📈 https://t.co/JvmhvVGmXl - **jeff_weinstein** (2026-05-19): https://x.com/jeff_weinstein/status/2056813179821527148 - Content owners can finally charge agents for accessing their site. Congrats Parallel on launching Index with such great partners! (All the micro money parts built using @mpp via @stripe.) https://t.co/2wi7qPK9MP - **grinich** (2026-05-19): https://x.com/grinich/status/2056812382056280413 - This is tonight! Come by for pizza and demos. 🔥 https://t.co/J08Hmvh3Od - **grinich** (2026-05-19): https://x.com/grinich/status/2056800050454634546 - 42% of these are powered by @WorkOS Many others making the switch to accelerate growth 🚀 https://t.co/oMkfXT48PQ https://t.co/LtIPjn0VD0 - **dinasaur_404** (2026-05-19): https://x.com/dinasaur_404/status/2056797856934109525 - another banger from the Artifacts team (@dillon_mulroy) you can now receive event notifications for repo changes push a commit -> kick off a review agent fork a repo -> kick off a build if you want access to the Artifacts beta, DM me 🙂 - **lalkaka** (2026-05-19): https://x.com/lalkaka/status/2056782680793166258 - effective LLM prompt caching makes or breaks inference budgets. tuning cache TTLs is hard and fiddly. we pointed persistent background agents at the task and saw a 77% reduction in cache write waste in a week. more on how here: https://t.co/e9niEKW2f5 - **CloudflareDev** (2026-05-19): https://x.com/CloudflareDev/status/2056757595482804595 - Cloudflare Tunnels 🤝 @AnthropicAI Claude Managed Agents https://t.co/0eeIFwo5BQ - **pheoru** (2026-05-19): https://x.com/pheoru/status/2056755520636059648 - @AnthropicAI just launched MCP tunnels for Claude Managed Agents. The networking layer underneath? Cloudflare Tunnel. AI agents can now securely reach MCP servers inside your private network — no inbound firewall rules, no public endpoints. Outbound-only, post-quantum encrypted. - **ritakozlov** (2026-05-19): https://x.com/ritakozlov/status/2056702679691239639 - very excited about this launch and proud to have @cloudflare play a part in both @claudeai announcements today looking forward to having more customers like @Amplitude_HQ build managed agents on @cloudflare + managed agents https://t.co/Fk1SNnDVko - **elithrar** (2026-05-19): https://x.com/elithrar/status/2056692332615889052 - > Zero-trust creds injection, custom egress proxies, better observability, private services via VPC speaking my language! https://t.co/T2eoJp4Ze9 - **MikeNomitch** (2026-05-19): https://x.com/MikeNomitch/status/2056659886126105051 - Excited to launch Claude Managed Agents on Cloudflare today! - Run sandboxes as microVMs or even lighter-weight isolates on CF - Zero-trust creds injection, custom egress proxies, better observability, private services via VPC - Agent Email + Browser Run + easy custom tool calls https://t.co/kqsdwYmBdl - **yuris** (2026-05-18): https://x.com/yuris/status/2056454506104438890 - “Mythos Preview is a real step forward, and it's worth saying that plainly before getting into anything else. We've been running models against our code for a while now, and the jump from what was possible with previous general-purpose frontier models to what Mythos Preview does today is not just a refinement of what came before.” - **darraghcurran** (2026-05-18): https://x.com/darraghcurran/status/2056429143395565605 - Fun to have my son join us at @fin_ai for work experience this week - so far he's demonstrated that he's a similar appetite to me during lunch AND having not written code yet in his life, built his first two games ever using Claude Code - AI feels like cheating to him because so far our school system mostly just reinforces that idea. - **jeff_weinstein** (2026-05-18): https://x.com/jeff_weinstein/status/2056424449721675833 - One of my favorite @stripe charts => Charging customers is one of the strongest possible signals to know if you're building something people want (to pay for). Excitingly, our data shows Atlas startups are doing so increasingly earlier in their startup lifecycle: nearly 1 in 4 companies get their first paying customer within 30 days of incorporation, up 3x since 2021. We'll keep a look out on the correlation of early monetization and finding product-market; we'll report back! - **dok2001** (2026-05-18): https://x.com/dok2001/status/2056393557418631292 - Mythos and other frontier models, pointed at live code across critical Cloudflare infrastructure. An honest read on what's working and what comes next. https://t.co/qvnMrX3gqW - **FredKSchott** (2026-05-18): https://x.com/FredKSchott/status/2056393549688537234 - hitting this interesting cross-roads with flue: 1) repo automation, workflows 2) hosted agents as the framework matures, the differences between them are becoming more obvious and more frustrating to design around (and by extension, for users). for example: in astro, it was a specific design goal that our repo automation and human maintainers would reuse 90% of the same content. Shared skills, tools, configuration, etc. etc. running "flue run triage" in a GitHub Action should be as close to a core maintainer opening up claude code in the repo and asking "triage this issue: URL" but if you're building and deploying a hosted agent, you want your skills and tools and subagents to live alongside the agent code, not the sandbox file-system. splitting your agent logic across "this logic (agent code, tools) lives in the codebase" vs. "this logic (skills, roles) lives in the sandbox" is a maintenance nightmare. i'm not sure what the answer is, but I see projects like Sandcastle by @mattpocockuk laser-focused on repo automation. I trust Matt to build something great here that will be hard for us to compete with. We are trying to do too much for too many people. meanwhile, I'm now talking with so many devs building agents (not just oss devs with oss repos) and there is no one doing what flue is doing today. A part of me really just wants to explore and optimize for this, and build the best framework for agents. idk, talking out loud a bit. will spend more time exploring this this week. curious if anyone who's tried flue (or considered it) has thoughts! - **dinasaur_404** (2026-05-18): https://x.com/dinasaur_404/status/2056388968686514565 - You can now use Artifacts directly from Wrangler! + dashboard support coming soon 👀 https://t.co/M1mvU66HTD - **threepointone** (2026-05-18): https://x.com/threepointone/status/2056319062595006594 - big +1. conceptually this removes a lot of questions around preview environments, staging urls, testing with "real" data, version skew, etc. you gotta be a little careful with database schemas and migrations, but you should already be doing that tbh. https://t.co/uGYYN13i7X - **tobi** (2026-05-18): https://x.com/tobi/status/2056198717225464307 - I’ve had very good results running autoresearch with local qwen 3.6 26b model as long as I had a simple vibed pi “advisor” extension that allowed it to periodically ask GPT 5.5 for ideas. I think this direction has a lot of merit. - **badlogicgames** (2026-05-17): https://x.com/badlogicgames/status/2056151438296850474 - recommended reading. https://t.co/IbM0wp93uF - **thdxr** (2026-05-17): https://x.com/thdxr/status/2056141119109443886 - ok tried the same on cloudflare for our x-rank tracking thing was just as easy and the built in sqlite is really nice, i was shoving stuff into blob storage on vercel https://t.co/Pj8mEpLb9m https://t.co/YJQlvyTZcp - **villi** (2026-05-17): https://x.com/villi/status/2056131797130661982 - The FDE conversation reminds me of software before SaaS. Sign an deal with SAP, buy a DB, servers and storage array boxes, hire an SI and a few consultants, and spend 18-24 months to get a solution that may not work. This is not the future. This is a sign of bad products. The future is good software that does not require FDEs to get to value. - **cnakazawa** (2026-05-17): https://x.com/cnakazawa/status/2055881241677668637 - Codiff 0.1 is out * Fast Local Code Reviews * Optional LLM Walkthroughs * Inline Review Comments This is the best companion for reviewing output of coding agents. macOS Release: https://t.co/wmySMLL1dK - **lessin** (2026-05-17): https://x.com/lessin/status/2055835145878528405 - Time for a re-read. https://t.co/mUCl0ppr7Y - **RhysSullivan** (2026-05-16): https://x.com/RhysSullivan/status/2055524681994883187 - i really wish the codex app was open source, so many small tweaks i want to make to it to better suit my workflow genuinely really well made piece of software it feels so smooth, i was unfamiliar with openai's game - **elithrar** (2026-05-15): https://x.com/elithrar/status/2055437259906679231 - very cool to see @tryhighlight building their global agent chat system on Workers + Durable Objects! https://t.co/DQmv8IGwLi - **RhysSullivan** (2026-05-15): https://x.com/RhysSullivan/status/2055412232729907389 - @steipete every time i see your workflow i realize how much i have to learn on building with these tools, genuinely always open my mind to how narrow in scope im thinking - **darraghcurran** (2026-05-15): https://x.com/darraghcurran/status/2055405875649503654 - how we AI https://t.co/cF7OjjHw5p - **RhysSullivan** (2026-05-15): https://x.com/RhysSullivan/status/2055378668613214216 - Excited to demo at MCP Night: Agent Mode! Come through it’s gonna be fun https://t.co/ryQwXVyyTC - **grinich** (2026-05-15): https://x.com/grinich/status/2055365850345918899 - MCP Night: AGENT MODE 🔥 feat. @RhysSullivan, founder of https://t.co/s2OqOgqei6 Happening next week. RSVP now: https://t.co/lXZVyle8RY https://t.co/UAU58r17xx - **fin_ai** (2026-05-15): https://x.com/fin_ai/status/2055348223162962310 - Most engineering orgs are experimenting with AI. We bet the entire company on it, and hit our 2x productivity goal 3 months early. Join us on May 19 for a live session where Fin CTO @darraghcurran, Principal Engineers @brian_scanlan and Kesha Mykhailov, and special guest @clairevo break down how Fin’s ~500-person R&D org: • Shifted to agent-first engineering • Got 95% of PRs running through Claude Code • Cut defects by 55% • Shipped 2x more product changes at nearly half the cycle time You'll hear about the messy middle nobody talks about – uneven adoption, code quality dips, the human bottlenecks that mattered more than the technical ones, and what they'd do differently. Register at the link in the replies. - **grinich** (2026-05-15): https://x.com/grinich/status/2055319904665174423 - MCP Night: AGENT MODE 🔥 feat. @ClaireVo, founder of @ChatPRD Happening next week. RSVP now: https://t.co/lXZVyleGHw https://t.co/ys1sdtFfgY - **insecureagents** (2026-05-15): https://x.com/insecureagents/status/2055311653521260879 - Thanks to all our panelists for a great discussion on building internal AI We learned how Cloudflare doubled MR velocity, Browserbase’s feature request pipeline runs at 100% coverage, and how Sentry turned a full debugging session into 1 slack message @pk_iv @zeeg @ianlivingstone @irvinebroque - **destraynor** (2026-05-15): https://x.com/destraynor/status/2055310868708274314 - Our new product, Operator, is prototypical of how all B2B software will be built. Say what you want and leave the rest to us. These 3 demos show case Operator doing analysis and synthesis, with dynamic UI, and inline collaboration. It's very, very, cool. https://t.co/C3rkh1l9O9 - **zebassembly** (2026-05-15): https://x.com/zebassembly/status/2055287789483172180 - @acoyfellow if this isn't screaming XAB idk what is - **acoyfellow** (2026-05-15): https://x.com/acoyfellow/status/2055235076820971872 - "capa" Turn OpenAPI specs into Cloudflare service bindings - Stripe, GitHub, GitLab, Jira, Slack, Twilio, Kubernetes, Box, Discord, Zoom + more - 13 capabilities so far, lmk what else to add - 5,852 generated methods you can bind to today https://t.co/66SU2tW8Qn https://t.co/pZRG87hrjp - **sperand_io** (2026-05-15): https://x.com/sperand_io/status/2055119008366760328 - If only there were a company with an ungodly infrastructure footprint and the product delivery bandwidth of all of yc combined Oh yea. $NET https://t.co/LQXKTJ6FQQ - **fin_ai** (2026-05-15): https://x.com/fin_ai/status/2055117294045388860 - We originally built the Fin Flywheel (Train, Test, Deploy, Analyze) to help humans get the most out of Fin. Now Operator can keep the Fin flywheel whirring. https://t.co/ANxRAaLO2N - **fin_ai** (2026-05-15): https://x.com/fin_ai/status/2055112576757887427 - What can Operator do? It can run & optimize every feature across @fin_ai and the @intercom helpdesk. It finds trends & opportunities and suggests how to take advantage of them. It's your expert in the system. https://t.co/XHp1ZzDzLI - **fin_ai** (2026-05-15): https://x.com/fin_ai/status/2055109298267566282 - In knowledge management alone, Operator compresses hours of work into minutes every time you use it. Operator will always serve you up suggested configuration changes to approve, so you're always in control. https://t.co/wjGRFu2IVe - **fin_ai** (2026-05-15): https://x.com/fin_ai/status/2055106410355064984 - Knowledge management a burden? Operator to the rescue. Watch Fin VP of Product @brian_donohue speed run a knowledge management task with Operator. https://t.co/w8nPEbHroF - **fin_ai** (2026-05-15): https://x.com/fin_ai/status/2055102980408418633 - So Fin, that's an agent for your customers. Operator, that's an agent for your customer operations. VP of Product @brian_donohue explains why Operator is a paradigm shift for how you interact with business software. https://t.co/BJ1E2AcbRj - **grinich** (2026-05-15): https://x.com/grinich/status/2055098475378061698 - patrick has been on an absolute heater the last few weeks with these guests one of the best podcasts out there today https://t.co/AQbsWff7dw - **tbpn** (2026-05-15): https://x.com/tbpn/status/2055089613870686521 - Raindrop cofounder @benhylak goes deeper on his exchange with @bchesky about booking Airbnbs with LLMs: "If Airbnb had an API, I would use it, and I would book Airbnbs through it using Claude Code. I know I would do it." "You see companies almost reducing themselves into an API with absolutely no moat. You look at Photoshop, Illustrator, etc. They're like, 'We have a Claude Code integration now.'" "At the point where people are just using Photoshop, Illustrator, etc. as an MCP, they've sort of lost the game, if no one is touching the UI anymore." "Right now, companies have to do that, increasingly, because they have no other choice. I think there will be a point where the incentives don't make sense anymore." - **dok2001** (2026-05-15): https://x.com/dok2001/status/2055084381916795139 - Congrats @Lovable on the ship. Powered by @Cloudflare Dynamic Workers. Each Lovable app gets its own Dynamic Worker, generated on demand, SSR'd at the edge. Exactly what this primitive was built for. https://t.co/2iiDmKNlSV - **kimmaicutler** (2026-05-15): https://x.com/kimmaicutler/status/2055077257845326000 - True. https://t.co/A6gTIx7gQC - **ianlivingstone** (2026-05-14): https://x.com/ianlivingstone/status/2054991847949570144 - Today, we’re proud to announce Keycard for Multi-Agent Apps. As agents begin coordinating with one another across applications, infrastructure, and clouds, the IAM model breaks down, leaving developers stitching together shared credentials, inherited permissions, and bespoke authorization logic. Keycard gives every agent its own identity, enabling delegated access across users, agents, and applications, with auth enforced at every hop in a single SDK & Control Plane. - **adityaag** (2026-05-14): https://x.com/adityaag/status/2054976557702066520 - https://t.co/AoyPtuI9jx - **dinasaur_404** (2026-05-14): https://x.com/dinasaur_404/status/2054954672138416166 - computa make the workers dash have a domains tab where i can buy domains ✨ you've been programmed ✨ https://t.co/jybyVXQxJj - **ritakozlov** (2026-05-14): https://x.com/ritakozlov/status/2054953372319375578 - developer platform, zero trust, security & performance. all in one cloud. developer platform, zero trust, security & performance. all in one cloud. developer platform, zero trust, security & performance. all in one cloud. are you getting it? https://t.co/xjxPEVOh1t https://t.co/9DV3AhvUR7 - **leinwand** (2026-05-14): https://x.com/leinwand/status/2054943005463982091 - 🚢🚢🚢 https://t.co/FUI2FTIPsP - **clairevo** (2026-05-14): https://x.com/clairevo/status/2054922108308992337 - On topic: there is a mistake I saw large companies make in late 2025/early 2026 which is they picked their coding model provider, chat tool, and harness and put the whole company on it. There was a big leap in model ability and folks felt confident. I have been walking into these teams and they have no idea how good codex is or that notion AI is sick af, of what’s what in the agent world. And now they’re sunk in a contract AND internal intertia while missing out on the best & most efficient tools of the moment. Let me say it again for people in the back: We are pre convergence on tools!!! Keep your organizational options open!!! Consumer (well, enterprise) choice is a gift when everyone is competing for your business and capabilities jump every month. Leverage it. - **utsengar** (2026-05-14): https://x.com/utsengar/status/2054915833496010821 - Rolling out support for Content-Signal header in robots.txt for all Webflow sites. Now, you can add: Content-Signal: ai-train=no, search=yes, ai-input=no PS: Still a proposed standard but Cloudflare has started to use it for their scoring in https://t.co/Wd4BQdqxBb https://t.co/BE7fZ9484w - **michellechen** (2026-05-14): https://x.com/michellechen/status/2054801966652326216 - we are an inference cloud (but not limited to) https://t.co/ALBdw9MxI4 - **bznotes** (2026-05-14): https://x.com/bznotes/status/2054791387636449299 - I remind myself of this quite often. https://t.co/l14iRcaNLX - **bryceadams** (2026-05-14): https://x.com/bryceadams/status/2054733210815512913 - Potentially useful tool from @Cloudflare to check how 'agent-ready' your site is https://t.co/aIKjtqmTqy. It took a few prompts but I managed to fix our score. https://t.co/tjDsWWEuoP - **threepointone** (2026-05-13): https://x.com/threepointone/status/2054598434942800097 - @dsp_ I had high hopes for sampling, too bad it didn't take off - **dinasaur_404** (2026-05-13): https://x.com/dinasaur_404/status/2054575739756785778 - working at cloudflare be like https://t.co/vXAEOoETSW - **villi** (2026-05-13): https://x.com/villi/status/2054395186877501644 - This guy is pretty awesome. Really excited for what he could accomplish as a governor. https://t.co/zisHvJU0zE - **moseskagan** (2026-05-13): https://x.com/moseskagan/status/2054388791675342896 - Sane Californians: Mahan is the guy https://t.co/BwMcxhRoCO - **kimmaicutler** (2026-05-13): https://x.com/kimmaicutler/status/2054377561346757023 - Woo woo 📢 https://t.co/NOvjyESxvv - **MattMahanSJ** (2026-05-13): https://x.com/MattMahanSJ/status/2054355868314169569 - California needs to move forward. @XavierBecerra needs to answer for his past. https://t.co/Hq4KRdHAbw - **lessin** (2026-05-12): https://x.com/lessin/status/2054338929978138907 - This made me LOL... ~8 years ago @kortina and I were working on the original Fin (as a data platform for optimizing customer service agents) & were in deep partnership conversations with Intercom. Today they become Fin (not us) -- we had the .com tho, not the .ai :) - **far33d** (2026-05-12): https://x.com/far33d/status/2054296146210906465 - Building stuff works. https://t.co/fm4lKBO2BP - **ezraklein** (2026-05-12): https://x.com/ezraklein/status/2054287610063786106 - Here's @MattMahanSJ, on the lessons — both politics and policy — of trying to cut unsheltered homelessness in San Jose to zero https://t.co/jkV3PdayDF - **Noahpinion** (2026-05-12): https://x.com/Noahpinion/status/2054283589634945305 - YIMBYism just works, period. Build market-rate housing supply. It works. https://t.co/NbPWcVvErV - **pitdesi** (2026-05-12): https://x.com/pitdesi/status/2054273860795850990 - So frustrating that Becerra is the frontrunner for governor in CA He was completely ineffective in the Biden administration, has bad policy ideas, and seemingly can’t take the heat. https://t.co/3nzJu4entH - **mnolangray** (2026-05-12): https://x.com/mnolangray/status/2054266328501141544 - For 45 years, Berkeley built virtually no new housing. By the mid-2010s, it was the most expensive college town in America. Shortly thereafter, YIMBYs took over and kicked off a building boom. Today, nominal rents are below 2018 rates—remarkable progress on affordability. https://t.co/EQrWsCG953 - **StartupArchive_** (2026-05-12): https://x.com/StartupArchive_/status/2054227600462201241 - Patrick Collison on the importance of waiting a really long time to hire people Everybody tells you “work really hard to hire the best people.” But as Patrick explains, that’s not helpful because everyone knows they should do that. As he puts it: “The question is to what length should you go and what does that actually mean in practice? And in practice, it means being okay waiting a really long time to hire people.” It took Stripe six months to hire their first two people, and six more months to hire another 3-4 people. He can think of five people at Stripe who took 3+ years to hire. “If you think about the smartest people you know, if you want to get them to work on your thing, chances are they already have pretty good paths ahead of them… You have to be way more persistent and be okay with it taking way longer than any sane or reasonable person would think it should take.” There’s an important compounding effect here — hiring just one great person will make it marginally easier to get the next great person. Patrick argues you should also view every person you hire as bringing along another 50 people just like them if your company is successful: “Even if they don’t literally hire 50 people, they will be so influential in determining the selection of those 50 people.” Source: @GreylockVC (Nov 2015) - **elithrar** (2026-05-12): https://x.com/elithrar/status/2054162732195197283 - pull_request_target has been one of the sharpest, most dangerous event types on GitHub. it's "necessary" to build/test fork PRs, but you're forever at risk of a malicious PR on a fork repo getting you. GitHub has warned against it for a long time. https://t.co/0gvtaPNXM2 - **thdxr** (2026-05-12): https://x.com/thdxr/status/2054078134211813846 - since compaction is good now i've been keeping a session pinned per PR im working on and reusing it until it merges what's nice about this is i can see the cost of the session and understand what the feature cost me (pinning sessions is under an experimental flag) https://t.co/213UZICkAJ - **ArmandDoma** (2026-05-11): https://x.com/ArmandDoma/status/2053983212465270989 - 74% approval for a big city American Mayor in his second year in office is crazy pants. Impressive stuff Mayor @DanielLurie https://t.co/2vPvZcenHh - **RhysSullivan** (2026-05-11): https://x.com/RhysSullivan/status/2053981313439605184 - @feross Dumb question but if socket is able to flag it in six minutes, why is Microsoft not? It’s crazy to me that they seem to be doing nothing here - **_bgiori** (2026-05-11): https://x.com/_bgiori/status/2053849193941475385 - Amazing visual guide to what makes flue different. https://t.co/S5HYBPuWXm - **yusukebe** (2026-05-11): https://x.com/yusukebe/status/2053758519002996871 - Everyone already may know, but Cloudflare is the best place to build AI Agents! Because we have the important 3: * Agents - Workers / Agents SDK * Model - Workers AI * Sandbox - Dynamic Workers / Containers https://t.co/Y9AER1RTcl - **dok2001** (2026-05-11): https://x.com/dok2001/status/2053680582547141006 - Built dozens of ad-hoc agent harnesses this year. All replaced by https://t.co/0J2HUKeg0e @momito's guide is a great place to start. https://t.co/P96BI8biMU - **credealjunkie** (2026-05-11): https://x.com/credealjunkie/status/2053670452443115674 - @BayAreaREMatt No more heterogeneous market in the US than Oakland - **dillon_mulroy** (2026-05-10): https://x.com/dillon_mulroy/status/2053617380052406354 - you love to see it. took the dogs for a walk and came back to results. artifacts test suite just went from ~1 minute down to 10 seconds with no changes to test coverage this would have taken me days to optimize and configure before ai https://t.co/550WQQSzcP - **hichaelmart** (2026-05-10): https://x.com/hichaelmart/status/2053612038593863795 - So we're talking: - Incredibly fast mock services (DBs, APIs, even browsers) - Incremental testing (including for E2E!) via dependency DAGs, etc - Parallel execution, local up to a point, then via remote sandboxes We'll want each agent iteration/experiment to take ms (ideally) - **momito** (2026-05-10): https://x.com/momito/status/2053573800814399838 - i’ve been playing around with flue, and have been really enjoying it i’ve created a visual guide, let me know what you think https://t.co/eU0vn8aNg7 - **levelsio** (2026-05-10): https://x.com/levelsio/status/2053471441174024552 - Oh I forgot to add, Cloudflare email is $5/mo (for Workers) but then includes 3000 free emails per month - **levelsio** (2026-05-10): https://x.com/levelsio/status/2053466448169685110 - ✉️ Trying @Cloudflare's new Email Sending feature today If you send 1,000,000 emails per month: - Postmark: $1,206/mo - Resend: $650/mo - SendGrid: $600/mo - Cloudflare: $354/mo - Amazon SES: $100/mo So Postmark is now by far the most expensive email provider And SES and Cloudflare are now the cheapest email providers I know my friend @marckohlbrugge is trying out SES now so I'll try Cloudflare and see how it is, SES is cheaper but Marc said it takes a bit more managing, and since I already use so much Cloudflare stuff it's nice to use them for email too With AI especially all of these are just as easy to use and setup in your app/site so economically it makes sense to go for the cheapest, because email is just email, it's all the same and deliverability is good with all of these I think TL;DR email sending has become a commodity! - **kr0der** (2026-05-10): https://x.com/kr0der/status/2053449022891753474 - cloudflare’s actually cooking i used to use it purely for domains but i’ve pretty much started using it + planetscale for everything, especially because AI agents make it very easy to get stuff set up - **AndyMasley** (2026-05-10): https://x.com/AndyMasley/status/2053296197679374365 - There are so many insane wildly misleading stories coming out about data centers almost every day now that I'm mostly having to give up on commenting on them to focus on actually getting blog posts out, but it feels like a tsunami. I'll share one from just today as an example. - **Tesla** (2026-05-10): https://x.com/Tesla/status/2053280404363432031 - The last Model S & the last Model X have been produced at Fremont Factory 14 years of history for Model S, 11 years for Model X 🫡 https://t.co/5sSscIe1f3 - **mattbaran** (2026-05-10): https://x.com/mattbaran/status/2053273302265999855 - Oakland is an unlivable hellhole. https://t.co/LT77hWlfmY - **evanjconrad** (2026-05-09): https://x.com/evanjconrad/status/2053238583545500064 - i dont think people understand that sf appears to be producing more millionaires per quarter than housing units - **RhysSullivan** (2026-05-09): https://x.com/RhysSullivan/status/2053238350648406378 - The Bun Zig -> Rust rewrite has me thinking about how important it is to have you tests in a separate language / method than you implementation This has always mattered but matters even more now. The reason they know the rewrite works is because nothing about the tests changed - **zachbruggeman** (2026-05-09): https://x.com/zachbruggeman/status/2053232909763629396 - Between us having built this at @tryramp with Inspect, and watching other great companies like @WorkOS, @stripe and @Shopify build this, some clear takeaways are emerging: 1. AI adoption multiplies exponentially when it’s done in public. If you work with tooling that keeps learnings private, you’re doing a disservice to your entire business. When every knowledge role is rebuilding how it works, you need everyone to contribute to the corpus of knowledge by default. 2. Bespoke tooling that’s shaped to your business is easier than ever to build. Losing time trying to shape your processes around other products isn’t a trade off you have to make anymore. Choose platforms that will let you stay flexible to build what works for you and your business. 3. Cultures of experimentation are more important than ever with AI. We are still so early. Shape your business to take big bets, and cut losses early. Whether this be for internal efforts like these, or the product you ship, now’s not the time to be risk adverse. It’s a far greater risk to think that anyone has won the game. - **addyosmani** (2026-05-09): https://x.com/addyosmani/status/2053231239721885918 - https://t.co/FSx0reCCLH - **eladgil** (2026-05-09): https://x.com/eladgil/status/2053206351158091819 - People at major AI labs (using internal models) 3-4 months ahead of startup silicon valley engineers SV founders/eng 3-6 months ahead of NY NY founders/eng 6-12 months ahead of rest of world Most people have no idea how fast AI shifting as 1-2 years behind SOTA "The future is here, just not equally distributed" - Robert Heinlein - **thdxr** (2026-05-09): https://x.com/thdxr/status/2053198657214632231 - @RhysSullivan unfortunately the timing of 5.5 coincided to when most of our codebase was ported to effect so i can't perfectly prove this but my agent use is up like 10x this month, getting incredible results, whole workflow has changed - **RhysSullivan** (2026-05-09): https://x.com/RhysSullivan/status/2053197707637113153 - With agents writing code I’ve become more bullish than ever on Effect and the problems it solves If you’re not using it, I’m curious on why and what you’re using instead to solve the same challenges - interested to see what the ecosystem looks like - **OpenRouter** (2026-05-09): https://x.com/OpenRouter/status/2053170520087024109 - Introducing Pareto Code: a new, free, experimental coding router Set `min_coding_score` in your request and route to the cheapest code-capable model that clears your bar, ranked by @ArtificialAnlys. See the Pareto frontier shifting in real time👇 https://t.co/26e1HeGOpp - **elithrar** (2026-05-09): https://x.com/elithrar/status/2053165943312818564 - love to see content like this building on top of the @cloudflare Agents SDK https://t.co/snivj6MYiz - **steipete** (2026-05-09): https://x.com/steipete/status/2053114837698249190 - Peekaboo 3.0 is live. Biggest release since 2.0. ⚡ Action-first macOS computer use 👁️ Unified screenshot + UI detection 🧩 Cleaner JSON across CLI + MCP 🛠️ Better snapshots I started this last year, but the models just weren’t good enough. Now they are. https://t.co/0wvhR0NWOj - **jarredsumner** (2026-05-09): https://x.com/jarredsumner/status/2053063524826620129 - there will be a blog post about this. on what this means for bun, benchmarks, memory usage, maintainability going forward, and also the literal process of doing this (it wasn’t just “claude, rewrite bun in rust. make no mistakes”) this is a 960,000 LOC rewrite, the code truly works, passing the test suite on Linux and soon other platforms. e2e I started working on this 6 days ago. this would’ve been a massive amount of work by hand. - **jarredsumner** (2026-05-09): https://x.com/jarredsumner/status/2053058171338682875 - why: I am so tired of worrying about & spending lots of time fixing memory leaks and crashes and stability issues. it would be so nice if the language provided more powerful tools for preventing these things. - **jarredsumner** (2026-05-09): https://x.com/jarredsumner/status/2053048478486708562 - it’s basically the same codebase except now we can have the compiler enforce the lifetimes of types and we get destructors when we want them. and the ugly parts look uglier (unsafe) which encourages refactoring. - **DhravyaShah** (2026-05-09): https://x.com/DhravyaShah/status/2052929981525971189 - Spotted in the wild. Insane 🫡🫡 https://t.co/ijvsb1TXwQ https://t.co/8Av7O8e9Dl - **dok2001** (2026-05-09): https://x.com/dok2001/status/2052917542029541820 - We finally convinced @eastdakota to let us do billboards in SF. I can’t wait to see them in person later this month. https://t.co/FHXvkSPBTX - **tbpn** (2026-05-08): https://x.com/tbpn/status/2052899342890201450 - .@bchesky says AI is the biggest opportunity for creative people in his lifetime. "I'm really worried that an entire generation of designers, artists, and creative people are going to decide to sit out AI." "The photo and video generation models allow you to design incredibly rich interfaces. This is the best time in the world for designers and creative people to get involved." - **momito** (2026-05-08): https://x.com/momito/status/2052880602441867547 - finally had a chance to play around with flue, and Fred cooked right level of abstraction and simple primitives that just make sense - v. delightful https://t.co/lL4Q5fX4qA - **DanielGri** (2026-05-08): https://x.com/DanielGri/status/2052866705530003843 - I am proud of the team that we do things like this. This is hard work and we make the world a better place. Also: Great example of effective use of AI to actually get more things done. https://t.co/Au8TNimAC7 - **ithilgore** (2026-05-08): https://x.com/ithilgore/status/2052843807809610078 - We’ve spent a lot of time on the framework underneath Codex, so it can move quickly on routine work while stopping for review when the risk changes. Here’s how we use sandboxing, approvals, network policy, and telemetry to run Codex safely @OpenAI: https://t.co/SXsYDARw40 - **OpenAI** (2026-05-08): https://x.com/OpenAI/status/2052800507727781979 - Just gonna leave this here. https://t.co/EOI980j9e9 https://t.co/HxbCl2Izcz - **destraynor** (2026-05-08): https://x.com/destraynor/status/2052778551884079505 - There is, of course, one AI Agent for Customer Support with public docs, self serve sign-up, public pricing, a CLI, API + more. It's the highest performing one, and we share all our ideas+research too. Product → fin .ai Research → fin .ai/research Ideas → ideas.fin .ai https://t.co/GGD6aVoCOo - **sinasanm** (2026-05-08): https://x.com/sinasanm/status/2052679431882465360 - kimiflare just crossed 10,000 downloads on npm 🎉 npm install -g kimiflare open-source terminal coding agent, built on kimi k2.6 + cloudflare. cloud version: get free 5M input tokens if you sign up before May 14 (no credit card needed). BYOK afterwards https://t.co/PceUGF34bz - **dok2001** (2026-05-08): https://x.com/dok2001/status/2052591562467492147 - @buccocapital That is a good piece. - **southpolesteve** (2026-05-08): https://x.com/southpolesteve/status/2052585203424522302 - @tannerlinsley credited vinext as part of the inspiration for his React s̶l̶o̶p̶f̶o̶r̶k̶ alternative projection https://t.co/rVj63MVJTR So I did the only logical thing. I slopforked his npm package and made it work with vinext: https://t.co/scktwEE98V The circle is complete. Full slopception - **stevehou** (2026-05-08): https://x.com/stevehou/status/2052578741906206952 - Beautifully written. One of the best I’ve read on the topic of the impact of AI on the workplace. Very thought-provoking. Highly recommended! 5/5 https://t.co/591Z7i2iAI - **buccocapital** (2026-05-08): https://x.com/buccocapital/status/2052566903399354657 - What’s actually happening is these businesses are stripping out the context carriers to hire more innovators and marketers. I wrote about what is happening here https://t.co/WS2NLM67Mv https://t.co/8ERemLk1Eb https://t.co/HwA5dOw4SJ - **eastdakota** (2026-05-08): https://x.com/eastdakota/status/2052560831909433554 - @taylorotwell Agree! Hence we laid off very few software engineers. Very few customer-facing sales people either. But if you don’t think AI changes how SalesOps, for instance, will be done going forward then it probably means you don’t know what SalesOps is. - **matthewcp** (2026-05-07): https://x.com/matthewcp/status/2052515665836142942 - Drop any Hono middleware into your Astro pipeline. Logging, auth, rate limits, wherever you want them. Astro's stages are just middleware now. https://t.co/iDikcYJxAl - **MattieTK** (2026-05-07): https://x.com/MattieTK/status/2052506959979376642 - You should update your Nextjs, OpenNext or any app using React Server Components. Cloudflare have been working over the last 24hrs to add as much protection as possible but there is no better protection than updating your app. https://t.co/MRKgTQK2oS https://t.co/h0rQ9wXfbk - **zabie_e** (2026-05-07): https://x.com/zabie_e/status/2052506086217846979 - Two engineers, same job, two LinkedIn job descriptions... Hire beaver on the left 10/10 times https://t.co/GaynmAb4Up - **CloudflareDev** (2026-05-07): https://x.com/CloudflareDev/status/2052467567588196703 - Multiple security vulnerabilities affecting React Server Components and Next.js have been disclosed. We strongly recommend updating your applications immediately. Cloudflare WAF managed rules already mitigate the disclosed denial-of-service vulnerabilities, and we are investigating additional coverage for several other CVEs. https://t.co/mT9ujk1H7c - **astrodotbuild** (2026-05-07): https://x.com/astrodotbuild/status/2052455580074131947 - Astro 6.3 is here! Experimental advanced routing lets you take full control of your request pipeline. Bring your own framework like Hono, compose handlers, and control exactly the order they run in. https://t.co/LzE2TDm2Pn - **nevikashah** (2026-05-07): https://x.com/nevikashah/status/2052448044243615769 - 🤸‍♀️You can now get a single unified trace of your Worker across service bindings and Durable Object calls 🤸‍♀️ no more manually piecing together multiple traces to see what’s happening within a request https://t.co/QsxjVchNuO - **mitsuhiko** (2026-05-07): https://x.com/mitsuhiko/status/2052387184519332094 - Look at this optimistic title. Fixed the blockers. In 469 files. https://t.co/BYC8MnprfE - **evanyou** (2026-05-07): https://x.com/evanyou/status/2052327497421946909 - It’s here! https://t.co/rw0KYeqfYo - **astrodotbuild** (2026-05-07): https://x.com/astrodotbuild/status/2052319416612761784 - Starlight 0.39 just dropped! New ways to use autogenerated sidebar links, plus improvements to styling and multilingual docs. Read all about it 👇 https://t.co/cZZhP3w1TI - **thsottiaux** (2026-05-07): https://x.com/thsottiaux/status/2052233392071360732 - Surprised there were basically no announcements today. - **FredKSchott** (2026-05-07): https://x.com/FredKSchott/status/2052207902124126463 - @jarredsumner bun add regrets - **insecureagents** (2026-05-06): https://x.com/insecureagents/status/2052090578347978868 - Everyone's building internal AI. A few teams are actually shipping it. Cloudflare nearly doubled MR velocity Browserbase cut support response to <24 hours Sentry turned 30-min investigations into 1 Slack message @zeeg @pk_iv @irvinebroque @ianlivingstone tell us how next week https://t.co/BmR1QtxVJW - **ritakozlov** (2026-05-06): https://x.com/ritakozlov/status/2052083779465019551 - built on @cloudflare sandboxes 🔥 https://t.co/f7yETFrlo6 https://t.co/CXmvGxzId7 - **jonas** (2026-05-06): https://x.com/jonas/status/2052078203687129268 - Watching gpt 5.5 use cloudflare's codemode MCP server and distributed tracing API to debug arcane worker subrequest exceeded errors feels like magic (Unfortunately this kind of debugging costs on the order of $100/hour to run at full pelt with high reasoning...) https://t.co/dQZMWOc6Lr - **kimmaicutler** (2026-05-06): https://x.com/kimmaicutler/status/2052066118869910012 - We're not going to be the party of affordability, sorry. https://t.co/taY4uedC8s - **BrandonMChu** (2026-05-06): https://x.com/BrandonMChu/status/2052048635093803310 - To most people, this seemed truly impossible back in the day @shopify is still underrated https://t.co/Ks2jeSTamY - **SidChatterjee** (2026-05-06): https://x.com/SidChatterjee/status/2051997378446520672 - Achievement unlocked Made it to the top of hacker news https://t.co/AQHlSbNWrr https://t.co/cfcndqzkv5 - **pitdesi** (2026-05-06): https://x.com/pitdesi/status/2051883298054652298 - Both the democratic frontrunners for governor want to kill more people by reversing the autonomous vehicle rules that the CA DMV already approved https://t.co/BOYJwTweTg https://t.co/yz60aISFfq - **thdxr** (2026-05-06): https://x.com/thdxr/status/2051874553152807111 - our teams last 7 days of spend damn gpt5.5 https://t.co/X0VPcUmRPm - **wesbos** (2026-05-05): https://x.com/wesbos/status/2051810722338750805 - Today I let my family down. I’ve worked for years, confident that whatever life threw at me I’d be able to tackle. If my family needs it, I’ll provide. Today my wife needed a cord. I confidently marched to my stash. 5v, 12, 19v 24v, even the obscure 27v. I have it all I was not ready for the curve ball life threw at me. 5v 6.5amp. Such high amps, so low volts. I came up empty. Cord guys I need your help. Was this all for nothing? - **kimmaicutler** (2026-05-05): https://x.com/kimmaicutler/status/2051790399879209119 - Willie Brown. https://t.co/xQk7CfPVVc - **MasterDotDev** (2026-05-05): https://x.com/MasterDotDev/status/2051699554878149026 - Most AI features ship as flaky demos. Scott Moss (Netflix) teaches the discipline of making them actually work he calls, "AI Engineering" Set up evals, and learn the improvement loop to make AI features actually behave! https://t.co/PZzpFKMILP - **jyangballin** (2026-05-05): https://x.com/jyangballin/status/2051677497562210552 - How much of SQLite, FFmpeg, PHP compiler can LMs code from scratch? Given just an executable and no starter code or internet access. Introducing ProgramBench: 200 rigorous, whole-repo generation tasks where models design, build, and ship a working program end to end. 🧵 https://t.co/8ayeDJLXaJ - **roerohan** (2026-05-04): https://x.com/roerohan/status/2051392122855858499 - Ran npx flue build --target cloudflare to see how it works. DurableObjects. Shocking. Love this framework though @FredKSchott - **jdevalk** (2026-05-04): https://x.com/jdevalk/status/2051351767393144864 - My blog scored 25/100 on @Cloudflare's https://t.co/xrdrGN5ycJ. A few hours later: 83. Wrote up the full stack: markdown content negotiation, Content Signals, Link headers, an API catalog, MCP, WebMCP. Plus where the scoring gets it wrong. https://t.co/icOwoPnWFB - **patrickc** (2026-05-04): https://x.com/patrickc/status/2051347945279332781 - I'm interested in "trapped buildings": those that couldn't be built today (because of zoning and code changes) but also can't be substantially modified or demolished (because of historic protection rules). One of those phenomena that really makes one wonder what exactly we're trying to do. Has anyone ever estimated what fraction of buildings in major cities fall into this category? When I asked Claude about San Francisco, it concluded: "If forced to give a single number with a single confidence rating: roughly 100,000 buildings — about two-thirds of San Francisco's physical structures — sit in the trap as a practical matter. Confidence: moderate. The number could be 70,000 or 130,000 depending on how strictly you operationalize "can't be substantially modified."" - **grinich** (2026-05-03): https://x.com/grinich/status/2051020455671128125 - 🎯 https://t.co/iTgsx41yJb - **dexhorthy** (2026-05-03): https://x.com/dexhorthy/status/2050931880971624880 - sign tapping will continue until software design improves https://t.co/RbiNEQSTxz - **thsottiaux** (2026-05-03): https://x.com/thsottiaux/status/2050811900062286168 - /goal might be the most consequential thing we have shipped in codex The value of good instructions has never been higher. - **addyosmani** (2026-05-02): https://x.com/addyosmani/status/2050720567473614968 - @dhh Modern JavaScript can totally be shipped #nobuild. When I was in Chrome, we experimented with how feasible this was at scale. You can load around 50-100 smaller scripts/chunks before the cost of IPC benefits from using a bundler/tooling. But you can go pretty far. - **eastdakota** (2026-05-02): https://x.com/eastdakota/status/2050709736580899164 - Built on @Cloudflare Workers. Peter is so cool! https://t.co/CMjMtywI8n - **atbeme** (2026-05-02): https://x.com/atbeme/status/2050683035650056454 - Let’s see what happens 👀 https://t.co/hgfLFpwA3D https://t.co/Hg2Nn5gi4H - **FredKSchott** (2026-05-02): https://x.com/FredKSchott/status/2050675029948723570 - Blown away by the response to https://t.co/oGh1BOOkkh over the last 24hr. Feels like we're onto something special here. Grateful to everyone who's tried it, shared it, or sent feedback. We're already on PR #35... 😅 - **sama** (2026-05-02): https://x.com/sama/status/2050671161915371998 - i keep thinking i want the models to be cheaper/faster more than i want them to be smarter but it seems that just being smarter is still the most important thing - **odysseus0z** (2026-05-02): https://x.com/odysseus0z/status/2050662967306547658 - @RamonGainez @FredKSchott @badlogicgames Try to implement exactly what that screenshot does in whatever tools you have right now - **CDTEliot** (2026-05-02): https://x.com/CDTEliot/status/2050658542999511105 - ed catmull basically solved one of the hardest organisational problems in creative work: how do you keep very high standards of taste without turning the people who hold that taste into gatekeepers that slowly strangle new ideas? his answer is elegant and I think underrated: you separate the people who set the bar from the people who make the final calls. - **dalmaer** (2026-05-02): https://x.com/dalmaer/status/2050656772860342317 - Fred and team always build such nice APIs https://t.co/9cRCSxTnlG - **michellechen** (2026-05-02): https://x.com/michellechen/status/2050654543487099090 - 24h later: workers ai is second cheapest effective pricing (behind moonshot), 78 TPS, 0.67 TTFT https://t.co/AIU98vkKXC - **odysseus0z** (2026-05-02): https://x.com/odysseus0z/status/2050654000534770078 - @RamonGainez @FredKSchott @badlogicgames Designed things I already do in spaghetti code so well that I can easily write them in clean code without losing flexibility - **DennisonBertram** (2026-05-02): https://x.com/DennisonBertram/status/2050637661375717579 - @odysseus0z @FredKSchott @badlogicgames I am really excited about it. Personally I went through the docs had Claude dissect the code and it looks great. - **threepointone** (2026-05-02): https://x.com/threepointone/status/2050605854383055317 - setup OoO responder for the first time in my life cheeky airport pint and I’m outta here https://t.co/sscroTj5in - **mrramibanna** (2026-05-02): https://x.com/mrramibanna/status/2050604173348270347 - I keep re-reading this masterpiece from @sidchatterjee & @irvinebroque - in awe of the explainer, and just how fast they moved to bring the Cloudflare services to the network. "Hey agent-of-choice, ask https://t.co/iftFXm0pQM to buy me a domain on Cloudflare. p&t" https://t.co/nEH3KK1jzk - **omooretweets** (2026-05-02): https://x.com/omooretweets/status/2050592759887900714 - OpenAI model release: We’re throwing a party 🎉 Everything is scribbles and Pets are in Codex. Hope you like goblins! Anthropic model release: In research preview, it hacked the full Internet for fun. Also, it’s coming for YOUR job specifically. Enjoy the permanent underclass! - **thdxr** (2026-05-02): https://x.com/thdxr/status/2050591210796937259 - american models are opaque there is no way to introspect anything about how they work or how they're hosted deepseek has released enough information that you can almost retrain their model on your own we're living in some weird upside down times - **thomas_ankcorn** (2026-05-02): https://x.com/thomas_ankcorn/status/2050528475446902896 - The local llm folks need to understand what this means. Shared kv caches make the economic reality of hosting your own model stupid. Just worked out the utilisation required to beat Cloudflares pricing for workers ai 😬 https://t.co/iXPb7RI9Ss - **thomasgauvin** (2026-05-02): https://x.com/thomasgauvin/status/2050404637425946685 - Been excited about this launch since I first saw it the agent frameworks are here, write once deploy anywhere/use any harness/switch to llms on the fly https://t.co/o3QOp3mk14 - **sama** (2026-05-02): https://x.com/sama/status/2050402088266694689 - /hatch clippy - **addyosmani** (2026-05-02): https://x.com/addyosmani/status/2050367119255327031 - @FredKSchott This is super well done, Fred! - **yenkel** (2026-05-01): https://x.com/yenkel/status/2050361209669460074 - looks very interesting yet another thing I’ll have to try out soon https://t.co/m0f3itRRt1 - **dannolan** (2026-05-01): https://x.com/dannolan/status/2050357408975982800 - @levelsio ERV - **FardeemM** (2026-05-01): https://x.com/FardeemM/status/2050353385694519570 - astro is goated. bullish https://t.co/waCJFijoFZ - **beffjezos** (2026-05-01): https://x.com/beffjezos/status/2050343213609701734 - Better message than Dario's "all your jobs are belong to us" https://t.co/R8GuLW2tCr https://t.co/SEgImKAxhR - **pmarca** (2026-05-01): https://x.com/pmarca/status/2050324118114070877 - Co-sign. https://t.co/KTKt4NYCZs - **thsottiaux** (2026-05-01): https://x.com/thsottiaux/status/2050316660519870937 - Please don't pronounce Codex, COD-ex. We are also not a fish. - **FredKSchott** (2026-05-01): https://x.com/FredKSchott/status/2050315629983572426 - 🥞 https://t.co/q6VpSm2SFr - **FredKSchott** (2026-05-01): https://x.com/FredKSchott/status/2050312935302688849 - Build your own vibecoding platform in 43 lines of code https://t.co/4AOleRGQQD https://t.co/snYKXOU7uG - **_bgiori** (2026-05-01): https://x.com/_bgiori/status/2050308429621932102 - If you're building sandboxed agents you should check this out! Wraps Pi-core and comes with out-of-the-box runtimes, and secret proxy security. I'm just happy I could be a part of the journey 😀 https://t.co/AnoTxus5y9 - **odysseus0z** (2026-05-01): https://x.com/odysseus0z/status/2050306516629258456 - Just to be clear, no model out there including Mythos whatnot has produced harness SDK/framework that is of the calibre of Flue by @FredKSchott and Pi by @badlogicgames. All of my current interaction with the software produced by the "tokens go burrr" school has significantly delayed my AGI timeline. They still all suck ass in terms of software architecture and you can just smell it everywhere. - **_bgiori** (2026-05-01): https://x.com/_bgiori/status/2050303994963325374 - @FredKSchott @FredKSchott Thanks for the shoutout! Flue is very, very easy to use. It's kind of crazy. - **burcs** (2026-05-01): https://x.com/burcs/status/2050301499058618763 - agent experience mentioned https://t.co/27NlERfg7M - **FredKSchott** (2026-05-01): https://x.com/FredKSchott/status/2050290387122872413 - @badlogicgames Not mad at all. - **badlogicgames** (2026-05-01): https://x.com/badlogicgames/status/2050289476895646007 - @FredKSchott how mad will you be if pi-coding-agent will have some of the abstractions found in flue? - **badlogicgames** (2026-05-01): https://x.com/badlogicgames/status/2050288716896247887 - ok, i already posted this but holy shit it's built on pi?! https://t.co/twV4rMAecs this makes me super happy! https://t.co/uAmUCDGMrk - **badlogicgames** (2026-05-01): https://x.com/badlogicgames/status/2050287678537535913 - this is really really nice and i will steal from it ruthlessly! https://t.co/uAmUCDGMrk - **FredKSchott** (2026-05-01): https://x.com/FredKSchott/status/2050287649928458619 - @tanishqk @threepointone Framework vs. SDK. Kind of like how Next.js/Astro is built on top of React, Flue is built on top of pi-agent-core to power our harness. Then we offer a different, more holistic DX on top of it. https://t.co/ODZPkMljcg - **meaganrgamache** (2026-05-01): https://x.com/meaganrgamache/status/2050276148140216360 - The biggest mistake I see new, early/mid-stage co VPs of Product make is thinking that they own the product and wanting to own product decisions entirely. This is not true. And, it's often detrimental to the company and the product. At the earliest stages, product and company strategy are inextricably linked. Any founder worth their salt is going to have a differentiated and unique point of view of their customer, market, and/or product. Your responsibility is to distill that point of view so that the team can understand and build with it in mind — and to refine it with customer feedback and data. Then, bring the founder and the team along with that feedback so that you can build an exceptional product and business. It's not easy or straightforward. It's a role that's primarily for someone who has exceptional judgement and ability to know when to lead and when to follow. - **FredKSchott** (2026-05-01): https://x.com/FredKSchott/status/2050274982232735812 - https://t.co/d73AbgeXIu - **FredKSchott** (2026-05-01): https://x.com/FredKSchott/status/2050274923852210397 - Introducing Flue — The First Agent Harness Framework Flue is a TypeScript framework for building the next generation of agents, designed around a built-in agent harness. Flue is like Claude Code, but 100% headless and programmable. There's no baked in assumption like requiring a human operator to function. No TUI. No GUI. Just TypeScript. But using Flue feels like using Claude Code. The agents you build act autonomously to solve problems and complete tasks. They require very little code to run. Most of the "logic" lives in Markdown: skills and context and AGENTS.md. Flue is like Astro or Next.js for agents (not surprising, given my background 🙃). It's not another AI SDK. It's a proper runtime-agnostic framework. Write once, build, and deploy your agents anywhere (Node.js, Cloudflare, GitHub Actions, GitLab CI/CD, etc). We originally built Flue to power AI workflows inside of the Astro GitHub repo. But then @_bgiori got his hands on it, and we realized that every agent needs a framework like Flue, not just us. Check it out! It's early, but I'm curious to hear what people think. Are agents ready for their library -> framework moment? - **meaganrgamache** (2026-05-01): https://x.com/meaganrgamache/status/2050268449667183077 - This is absolutely true. Consumer PM is an exceptional training ground for B2B — as well as dev tools, for the right kind of PM. https://t.co/YbsHZHigOZ - **ritakozlov** (2026-05-01): https://x.com/ritakozlov/status/2050247030463693213 - today is a good day to try kimi k2.6 on @cloudflare workers ai https://t.co/6qnYouXX3Q - **michellechen** (2026-05-01): https://x.com/michellechen/status/2050245669936378020 - we just made kimi 2.6 even faster and cheaper on workers ai. happy building! https://t.co/SepyK09yLs - **rwdaigle** (2026-05-01): https://x.com/rwdaigle/status/2050245323755225454 - Just had an amazing @Cloudflare agent experience In their admin dashboard I was able to fully purchase, provision, and configure services via the help agent An actually empowered agent that's not just a search bot They get it - **WorkOS** (2026-05-01): https://x.com/WorkOS/status/2050240254766944672 - Our blog bot wrote this post. It also pitched the topic itself. @Cloudflare Durable Workflows turned what would've been queues + workers + callbacks + checkpointing into step functions. A multi-model AI pipeline runs durably behind every Slack message, and our team is now publishing like crazy - getting engineering stories and announcements out in minutes that previously would have sat in the backlog for weeks. https://t.co/vu2qDl8PdA - **dinasaur_404** (2026-05-01): https://x.com/dinasaur_404/status/2050239803522695480 - dynamic workers x workflows is out! perfect for agents that need to execute tasks, run long-running workflows that survive crashes, sleep for days, and pick back up exactly where they left off the platform defines the workflow handling in the loader and the dynamic worker just calls env.WORKFLOWS.create() like any other binding. under the hood that's just RPC back across the dynamic worker boundary. you can use the same pattern to build entirely custom bindings of your own! - **kevin_flansburg** (2026-05-01): https://x.com/kevin_flansburg/status/2050238819065299110 - Yesterday Workers AI rolled out two major changes for Kimi k2.6: speculative decoding and shared KV cache. The first should improve TPS, especially for predictable output like tool calls. The second should improve cache hit rates, lowering TTFT and blended cost. https://t.co/Tl7LwhYpRm - **dillon_mulroy** (2026-05-01): https://x.com/dillon_mulroy/status/2050229247222956088 - artifacts dashboard coming soon™️ https://t.co/SBLA3NHRcM - **sama** (2026-05-01): https://x.com/sama/status/2050229058425045178 - we want to build tools to augment and elevate people, not entities to replace them. - **badlogicgames** (2026-05-01): https://x.com/badlogicgames/status/2050225380640882995 - really like this anthropic podcast. learning so much new stuff about anthropic every episode. it's a good anthropic podcast. https://t.co/OlBooHHHoN - **celso** (2026-05-01): https://x.com/celso/status/2050211184129786084 - Agents can now add durable execution to their plans with Dynamic Workflows. https://t.co/rDt602n5Dg - **patrickc** (2026-05-01): https://x.com/patrickc/status/2050206647793144098 - Stripe Atlas just hit 100,000 all-time incorporations. Q1 2026 is +130% Y/Y. https://t.co/ea5OQyjp0h - **schrockn** (2026-05-01): https://x.com/schrockn/status/2050200001322557877 - I've seen enough: churning from claude to pi. What a minimal, elegant, powerful piece of software. Malleable software with well-structured foundations is going to spread like wildfire. Pi is the showing us all the way. - **colinhacks** (2026-05-01): https://x.com/colinhacks/status/2050037441856504021 - happy Node 20 EOL day. a perfect day to mention: i plan to drop CommonJS in the next minor version of Zod - avoids needless ecosystem churn from a new major - all LTS Node versions support require(esm) - no new major releases of Zod are planned atm - **sama** (2026-05-01): https://x.com/sama/status/2050021650641695108 - artificial goblin intelligence achieved - **kentcdodds** (2026-04-30): https://x.com/kentcdodds/status/2050000659865575897 - Really starting to get irritated by services I use not being agent ready. I'm not talking about MCP necessarily. I'm talking about not even having an API of any kind. So irritating! - **dillon_mulroy** (2026-04-30): https://x.com/dillon_mulroy/status/2049991673850282366 - forking company gateway support lfgggg https://t.co/SZu7b6THLV - **BenjDicken** (2026-04-30): https://x.com/BenjDicken/status/2049989374646043084 - I don't need more software. Would appreciate the software I already have to work better, though. - **badlogicgames** (2026-04-30): https://x.com/badlogicgames/status/2049988700432404829 - People of https://t.co/oUoqqL9hAp New @MistralAI model, and forking company AI Gateway release! I also ripped out: - Antigravity and Gemini CLI support - mom, see https://t.co/El4j7mtl61 instead, build your own, or use an old commit - pods I'll never get to pi server ... https://t.co/Jv5UMvM9MN - **jeff_weinstein** (2026-04-30): https://x.com/jeff_weinstein/status/2049987111701606461 - if you want to see true flow state, watch @stevekaliski demo machine payments. https://t.co/vwQlhbQo1G - **maria_rcks** (2026-04-30): https://x.com/maria_rcks/status/2049986615217418694 - people will see: - tibo reseting limits daily - sam being a real human and interacting with people (and getting some dunks on) - the codex team always talking to people who use codex - giving subs to open source projects & open sourcing their harness and ask "why are you guys so nice to openai" - **stevekaliski** (2026-04-30): https://x.com/stevekaliski/status/2049950557989995005 - I joined @StripeDev at Stripe Sessions to chat all things @link cli and @mpp --> https://t.co/0zwRQ01Q9m - **thdxr** (2026-04-30): https://x.com/thdxr/status/2049946137856020788 - the whole "cannot use our models to develop a competing product" is getting really confusing now that model labs are launching regular products like Claude Design - **ritakozlov** (2026-04-30): https://x.com/ritakozlov/status/2049941719551213817 - a customer went from 0 to 1M dynamic workers spinning up daily in less than a week imagine doing that kind of scale up with any other type of sandbox - **agarwal** (2026-04-30): https://x.com/agarwal/status/2049927456187146537 - I’ve been using Casa for a couple of months and I’m very impressed. Strong recommendation! https://t.co/YgToVlDAEi - **zachbruggeman** (2026-04-30): https://x.com/zachbruggeman/status/2049912136957386848 - When we first shared Inspect, @tryramp's in-house coding agent, it wrote 30% of all merged PRs. Today, it's ~70% of all merged PRs, and goes far beyond just our engineers. We shared some successes with the Ramp team today in our all-hands, that I wanted to share here as well. https://t.co/pjvwT0tgNS - **kayvz** (2026-04-30): https://x.com/kayvz/status/2049908010009706663 - michael is a special founder and @getcasa is an incredibly special product. it's one of those "one way door" experiences where once you use, you cannot imagine being a homeowner *without* having Casa. if you live in SF or LA and own a home, you need to sign up. https://t.co/bsB2tPdAt4 - **gakonst** (2026-04-30): https://x.com/gakonst/status/2049891697015087275 - pretty incredible to see @patrickc casually mentioning to @sama the agentic infra we built in January for @tempo and @paradigm to transform how we work. we indeed use this to orchestrate pretty much anything at both companies. neither paradigm nor tempo can live without this now. it feels like the stone age without it. and because our organizations are so workflow heavy / data rich and we are <100 people, these are the perfect organizations to max deploy ai. it is natively multiplayer, harness agnostic, has secure secret management & firewalls, durable workflows, observable so it can self-improve/heal and most importantly - all self-hosted on bare metal & self-built, end to end. no third parties involved. we have two deploys, one for each company on different machines, but they both have a shared core codebase that's modular and extensible for each company's idiosyncracies. we're looking to do way more things with it. proactive agents remains an open problem for us. we hope to open source it very soon as i've mentioned a few times before. we're adding 1 special eng to that team, dm me + your proof of work if you're epic at infrastructure / security and love hacking things quickly. - **getcasa** (2026-04-30): https://x.com/getcasa/status/2049873696484143397 - After two years of building under wraps, today we're announcing Casa – your personal property manager. We've raised $27M to redefine the homeownership experience from the ground up. We believe your home is your most treasured asset, emotionally & financially. It shouldn't also be a second job. Most homeowners are on their own – expected to have the time, expertise, and relationships to keep things running. Finding a plumber you can trust. Remembering when the HVAC was last serviced. Knowing what's actually wrong before someone shows up to fix it. Casa gives every homeowner what used to be reserved for the few: a dedicated team that knows your home deeply, handles the work, and stays in your corner. We're enabling this by building a deep, technical understanding of every home we serve – something that's never existed before, across 100 million single-family homes in the country. For $199/mo, membership includes: - A complete inventory of your home, built using specialized hardware & software - 1.5 hours of handyman time every month (and it rolls over) - Unlimited Concierge requests to take on virtually any home project - Custom, proactive care plans built specifically for your home - Weekly package and donation pickups - Scheduling and payments for your regular vendors - Utility and property tax monitoring …and we’re just getting started, with more benefits on the way to make the experience of owning your home as magical as it always should have been. Available now in the SF Bay Area and Los Angeles. Reserve your spot everywhere else. → https://t.co/vTUSR000fC - **ashleyhindle** (2026-04-30): https://x.com/ashleyhindle/status/2049870559874286039 - 👋 Vask says howdy: Websockets, but better 🔥 • Powered by Cloudflare • No fanout tax: 1 broadcast to 500k = 1 message • Drop-in Pusher replacement First 20 peeps: code ASHLEY-ISNT-THAT-BAD = 50% off forever (only if you agree, Stripe checks) https://t.co/L2kbFeHyOb - **Cloudflare** (2026-04-30): https://x.com/Cloudflare/status/2049869781105353057 - We have recently added "Docs for agents" in our docs homepage to help you use your favourite AI agents to consume Cloudflare documentation. https://t.co/9cJ5ehjX6K https://t.co/ir4s13q0CD - **mrramibanna** (2026-04-30): https://x.com/mrramibanna/status/2049853493104836711 - https://t.co/9o3d7oNWgO - **threepointone** (2026-04-30): https://x.com/threepointone/status/2049850243701764190 - been working on subagents etc for the last few days, and it's come along well. some highlights: (tldr - we have the world's best subagents. they're just baby durable objects.) - subagents as separate threads so you can host all of a user's/entity's chats in a single DO (examples/multi-ai-chat) - subagents now have a first class story as "tools", with resumability, recovery, persistence etc all built in. works with ai-chat and think. (examples/agents-as-tools) - a new example showing how to use github auth for your agents (because I want to actually deploy something live for my friends and I to use. examples/auth-agent) - can now configure beforeStep, reasoning, telemetry for think. - fixed a few nasty timing/consistency bugs just did a release. now starting a big docs sync for everything we shipped in the last month, aiming for it all to go out tomorrow. we should update the starter with the new fun stuff as well sometime... npm i agents - **badlogicgames** (2026-04-30): https://x.com/badlogicgames/status/2049757619150004286 - look at me. i'm the harness now. https://t.co/Ff4Km8iQ3I - **ivanleomk** (2026-04-30): https://x.com/ivanleomk/status/2049733178462773317 - 1. @Cloudflare tunnels are great for providing a ngrok equivalent for remote agents 2. @GoogleDeepMind models are fantastic for computer use 3. Profit with remote browser use in a few steps hehe once u hook up @__morse's playwrighter though it did take some mods to work - **sama** (2026-04-30): https://x.com/sama/status/2049715178611380317 - alignment failure https://t.co/jFquokO1MK - **garrytan** (2026-04-30): https://x.com/garrytan/status/2049713659103129810 - Anthropic shouldn’t do this and doesn’t have to Would be great if it stopped https://t.co/DBUoPUpqnW - **patrickc** (2026-04-30): https://x.com/patrickc/status/2049705418436600244 - We just announced a large raft of improvements at @Stripe Sessions. My meta reflections: • It feels that the entire economy is replatforming right now. • Many charts at Stripe are inflecting in quite dramatic ways. What GitHub recently reported for commits we are seeing in economic activity (such as new company formations). • It is increasingly clear that agents will be responsible for most transactions in the not overly distant future. • Stripe was always developer-centric, but AI is making developer-centricity strategic in a new way: agents are even hungrier for good DX than developers themselves are. • Things that we’re launching are increasingly network products at heart. (Instant transfers between Stripe businesses, new kinds of fraud prevention with Stripe Radar, stablecoin payouts to anyone with Link.) "How can we turn Stripe's economies of scale into user benefits?" is increasingly the relevant question. • Between Privy, Bridge, Tempo, and Stripe’s core capabilities, we’re now doing a lot in stablecoins/crypto, and companies like DoorDash, Ramp, Meta, and Klarna are using our crypto stack to deploy meaningful new functionality in production. “But where’s the production use?” is rapidly becoming stale when applied to crypto. • After more than a decade of building, we seem to have hit some kind of critical mass of core platform capabilities such that building new things now feels easier and faster than before. (AI also helps.) We announced Stripe Treasury last year (originally called Financial Accounts); since then, we’ve added multi-currency support, global payouts, card issuance and rewards, and a bunch of other sophisticated functionality. By the end of this year, Treasury will support 15 more currencies and be available to businesses in 160 countries. On the launches themselves, a small selection that I thought were cool, though this is really just a subset: • The @Link AI wallet. Point your agent to https://t.co/vYdvNtJgpE and ask it to make purchases on your behalf with secure single-use tokens. (To test it, I asked Claude Code to buy a small gift for me yesterday. It purchased HTTPZine on Gumroad.) • New payment methods for Link, including Pix (largest payment method in Brazil) and UPI (largest payment method in India). We’re also adding stablecoin support to Link (which I think will be huge if we execute well). • We’re adding a lot of new Machine Payments Protocol functionality, including micropayment and recurring payment support. • We announced Checkout studio: a sophisticated dashboard for managing your checkout flow, including things like transaction replays and A/B tests. Today this tends to require a lot of fussy edits to production code. • Adaptive Pricing (which automatically localizes the price and currency that customers see) now supports subscriptions. We’ve seen pretty huge (4–5%) conversion rate improvements after enabling it — customers really like paying in their home currency. • New Stripe Terminal reader (the T600) with a customer-facing screen that can run native apps, plus support for 15 new international markets for Stripe Terminal. • General availability for Stripe Managed Payments, our merchant of record solution. (Natively handles tax, disputes, fraud.) Maybe sounds a bit arcane, but it’s one of those iykyk products. It saves a lot of schlep. • Fraud is a *much* bigger priority for customers than it was 2 years ago (AI makes fraud easier + unlike software, tokens can be resold), so we’ve been extending Stripe Radar to support things beyond payments fraud: free trial abuse, multi-account abuse, pay-as-you-go abuse. Early results are extremely positive. We also announced Stripe Signals — new scoring APIs for customers, businesses, and other objects, not just payments on and off Stripe. • Usage-based billing is also becoming the de facto business model of the AI era, and we launched a bunch of new pricing models in @getMetronome and features like low-balance alerts, automatic credit top-ups, and multidimensional pricing structures. • We showed streaming payments built on @Tempo and Metronome — track usage and get paid the instant value is delivered. Hard to predict, but I think this could be big. (Why wouldn’t you want to get paid as costs are incurred?) • We added automatic US tax filing in Stripe Tax. • We announced Stripe Database -- a hosted PostgreSQL database with all of your Stripe data, updated in real time. Read-only to start but we’ll make it read-write. • Stripe Workflows are now GA. • We showed Stripe Console, a full agentic execution environment built directly into the Stripe Dashboard. It’ll happily write code and use tools to answer your questions. • We previewed custom objects: model your business data directly in Stripe, with custom objects, typed fields, and relationships. • As mentioned above, Stripe Treasury accounts will support storage in 15 currencies by the end of the year. And instant/free(!) transfers between US Stripe businesses. • You can use a Stripe card with your Treasury balance and get 2% cash back on purchases. • We’re massively expanding our Global Payouts coverage -- soon 100 countries with fiat rails and 160 with stablecoins. • Atlas companies can now raise money directly within Stripe. • We launched the platform growth studio, which uses Stripe’s network data to generate specific recommendations for optimization/growth. • We announced the Stripe Managed Risk API — platforms can outsource risk handling to Stripe while maintaining full UI/UX control. • Connected accounts now benefit from networked onboarding, which hugely increases conversion rates. • We’re launching Treasury for Platforms. Connected accounts can get spend cards with just a few lines of code. (Plus cash rewards, cash acceptance, check acceptance, real-time payments…) • We announced Issuing for agents: easily create cards for agents. But that’s really just a subset of a subset. (See https://t.co/Ej0S8aRVi0 for more.) The Stripe team is cooking! And if you’re interested in building the economic infrastructure for this new world, we’re hiring. - **TheRealAdamG** (2026-04-30): https://x.com/TheRealAdamG/status/2049702640255201745 - https://t.co/h4QJVmYQNu “In this post, we'll explain how we made agent loops using the API 40% faster end-to-end, letting users experience the jump in inference speed from 65 to nearly 1,000 tokens per second. We approached this through caching, eliminating unnecessary network hops, improving our safety stack to quickly flag issues, and—most importantly—building a way to create a persistent connection to the Responses API, instead of having to make a series of synchronous API calls.” - **sherwinwu** (2026-04-30): https://x.com/sherwinwu/status/2049700835916210362 - One of my favorite of our recent blog posts: @cerebras made GPT‑5.3‑Codex‑Spark so fast that we had to rethink how Codex uses the Responses API – leading us to build WebSocket support for ultra fast latency. Cerebras speed is just 🤯 https://t.co/9AKuqXbNRz - **Laurentia___** (2026-04-30): https://x.com/Laurentia___/status/2049699622965387593 - Never did I imagine a feature I worked on would lead to goblin amplification, but here I am. Happy reading to those of you who are goblin-curious. https://t.co/1AXi0ApFk3 - **thdxr** (2026-04-30): https://x.com/thdxr/status/2049694348095148410 - tired of this misinformation so we made a video on the truth behind the anthropic vs opencode drama https://t.co/Vl3J2XZ9DX - **sama** (2026-04-30): https://x.com/sama/status/2049692014586048973 - https://t.co/7Am81lOEP3 - **sama** (2026-04-30): https://x.com/sama/status/2049691999444639872 - goblinblog dropped - **thsottiaux** (2026-04-30): https://x.com/thsottiaux/status/2049690702452527311 - Never talk about goblins. Our latest blog is live. https://t.co/r39VCx7TYZ - **GergelyOrosz** (2026-04-30): https://x.com/GergelyOrosz/status/2049690661201252630 - The last ~month, Anthropic really behaves like it was bought by private equity: 1. Raises prices without offering anything meaningful in return (Opus 4.7 tokenization) 2. Poor customer support for paying customer (silent bans of a few 50+ seat corp accounts) 3. Shady growth hacks that go against the company's previously stated values on integrity (eg silent updates on the pricing page, removing Claude Code from $20/month packages) - **OpenAI** (2026-04-30): https://x.com/OpenAI/status/2049690533845663830 - We’re talking about Goblins. https://t.co/dqmcLGCW71 - **Tom_A_Lynch** (2026-04-30): https://x.com/Tom_A_Lynch/status/2049687470741098886 - Way cool The machines will a be ubiquitous economic force https://t.co/ISj7qcvYw2 - **thsottiaux** (2026-04-30): https://x.com/thsottiaux/status/2049684345766744401 - A lot about OpenAI can be understood by realizing that a lot of us believe that we can at the same time deeply care, do the best work of our lives AND have fun too. No goblins to see here. - **dani_avila7** (2026-04-30): https://x.com/dani_avila7/status/2049665684146151683 - OK Cloudflare just took my weekend Going to try something, use Claude Managed Agent with this new Cloudflare + Stripe Projects flow and see how far it actually gets on its own Real end-to-end experiment, no shortcuts I'll post what happens, whether the agent really pulls off everything the blog claims - Create the Cloudflare account - Register the domain - Deploy to production All without me touching the dashboard 🧐 - **draecomino** (2026-04-30): https://x.com/draecomino/status/2049656387014443200 - Cloudflare is the most live player of all the compute clouds. They are the first to let the AI agent be a billable customer. https://t.co/BCQ0g8Gn7e - **sama** (2026-04-30): https://x.com/sama/status/2049653810558353746 - GPT-5.5 is going to have a party for itself. it chose 5/5 at 5:55 pm for the date and time. if you'd like to come, let us know here: https://t.co/OupLcJnf14 codex will help the team pick people from the replies. 5.5 had some good ideas/requests for the party, which we'll do. - **theo** (2026-04-30): https://x.com/theo/status/2049653251004662141 - A letter to my friends at Anthropic I hate that I feel obligated to do this. I hate that I've had to be so harsh towards Anthropic for the past few months. I really, really don't want to. I know it might feel like I'm doing this for clicks or something, but I promise I'm not. My pro-Anthropic content ALWAYS outperforms my anti-Anthropic content. I have cost myself a lot of money, opportunities, sponsors, and more. I'm doing this because you work for an evil cult. I'm begging you to wake up. Your CEO, Dario, does not respect engineers. This is obvious. He couldn't make it more obvious if he tried (and I think he's trying pretty hard) You know this, but you don't want to acknowledge it. It has kept you up many nights. You know that bad code is shipping to users. You know that one bad tweet might get you fired. You fear for your vesting schedules. You're afraid. Nobody deserves what you're going through right now. You go to work afraid, you leave work afraid, and you go to YouTube to keep up on the dev world, just to hear me yelling all about how evil your company is. You deserve better. You might not feel like you do, but you know deep down that this isn't right. I hope you know how deeply I feel for you. I'm sorry. I know I haven't helped you much individually, and I want to be better about this. If you're ready to leave, please hit me up. I swear I'll never tell a soul. I have friends at every lab and most startups in the AI world. Most of them would be down to match your current vesting schedules, possibly even go beyond. If you're staying for the money, I beg you to hit me up. We can make the money happen somewhere that hates you less. I know I'm asking for a lot of trust here, and that you're scared after seeing how hard I've been on Anthropic. I can't blame you at all for that. I should have posted something like this months ago. That's my failure to own and I will own it to my best ability. If you're willing to trust me in this moment, I can make it right. Let me help you escape. You deserve to work somewhere that you can have impact. Somewhere that listens when you feel something is wrong. Somewhere that won't fire you when you point out the things that hurt your users. My DMs are always open to you. When you're ready, let me know. I promise to make it right. - **VibecodrSpace** (2026-04-30): https://x.com/VibecodrSpace/status/2049652406422749252 - @_ashleypeacock @aarondfrancis A durable object is anything you need it to be. An actor. A room. A user's session. A document. A game lobby. Compute and state in one place, one writer, globally addressable. Aaron's series makes that click. - **theo** (2026-04-30): https://x.com/theo/status/2049645973350363168 - Fun fact - if you have a recent commit that mentions OpenClaw in a json blob, Claude Code will either refuse your request or bill you extra money. This is an empty repo, I'm just calling Claude Code directly. Insanity. https://t.co/JKthsztu1X - **mrmagan_** (2026-04-29): https://x.com/mrmagan_/status/2049626517769240606 - @irvinebroque i built a bunch of skills to automate provisioning of resources, including cf resources. even those skills got stuck on what could be done over cli or even api. domains still required many manual steps... - **EastlondonDev** (2026-04-29): https://x.com/EastlondonDev/status/2049621403054317890 - So cool https://t.co/pQ8lJoDEpn - **morganlinton** (2026-04-29): https://x.com/morganlinton/status/2049598053427109947 - @Cloudflare Love this - **MattieTK** (2026-04-29): https://x.com/MattieTK/status/2049597641986568233 - @caiobchi @Cloudflare @CloudflareDev hey! how are you managing the jekyll site? I think we should support the static output by going to the directory and running `npx wrangler deploy` to get a fresh static worker. - **paulg** (2026-04-29): https://x.com/paulg/status/2049593723390976457 - @cb_doge I've already explained that this "quote" from me is bullshit. But the really striking thing here is how similar this tweet is to other ones appearing just as Musk v. Altman is happening. I smell a coordinated PR campaign, and I'm not going to be used as part of it. - **darkmembo** (2026-04-29): https://x.com/darkmembo/status/2049587375542608073 - if i tell you that cloudflare is a wild place, i promise you do not understand. - **svensauleau** (2026-04-29): https://x.com/svensauleau/status/2049581553374679403 - Getting so much easier to start your next side project and forget about it a few days later https://t.co/4qM14G5mZb - **sarthakgh** (2026-04-29): https://x.com/sarthakgh/status/2049574494449303797 - Only 288 product updates today https://t.co/YIHBQvwcGu - **dillon_mulroy** (2026-04-29): https://x.com/dillon_mulroy/status/2049570435096858699 - i hate flakey tests i hate flakey tests i hate flakey tests i hate flakey tests i hate flakey tests i hate flakey tests i hate flakey tests i hate flakey tests i hate flakey tests i hate flakey tests i hate flakey tests i hate flakey tests - **dillon_mulroy** (2026-04-29): https://x.com/dillon_mulroy/status/2049565600876581121 - effect on cloudflare is so back. already can tell alchemy will be the way i build https://t.co/YDdSrOlK01 - **mrramibanna** (2026-04-29): https://x.com/mrramibanna/status/2049560806749343930 - @garrytan you ask, we delivered https://t.co/VfpOD5ORhC - **mrramibanna** (2026-04-29): https://x.com/mrramibanna/status/2049560257870135782 - incredible demo. purchasing a domain never been this easy or agent’y. Cloudflare absolutely cooked with this… https://t.co/jKIfdUh8FB - **kentcdodds** (2026-04-29): https://x.com/kentcdodds/status/2049557873248940061 - @Cloudflare @threepointone 👀‼️ - **jeff_weinstein** (2026-04-29): https://x.com/jeff_weinstein/status/2049556893954453572 - tokens are the atomic unit of an ai-native biz model. with @stripe you can: (1) protect margins by auto-adjusting prices to include latest model costs (2) get paid for token consumption in real time with @metronome and @tempo (3) stop pay-as-you-go token abuse with Stripe Radar - **jeff_weinstein** (2026-04-29): https://x.com/jeff_weinstein/status/2049556846084841713 - my favorite thing we’ve ever shipped! @link is now an easy way to permission agents to pay on your behalf. https://t.co/33E0i8UGZy - **dwr** (2026-04-29): https://x.com/dwr/status/2049552043845857610 - Add MPP to any API with Cloudflare https://t.co/XabTKzaqQd - **jeff_weinstein** (2026-04-29): https://x.com/jeff_weinstein/status/2049551340414914918 - When we first launched the @stripe CLI seven years ago, humans kinda liked it. Today, agents *love* it. https://t.co/0Fn0FLDuXh - **grinich** (2026-04-29): https://x.com/grinich/status/2049550896279998727 - We are thrilled to be part of the Stripe Projects launch 🥳 Try it today. Just run > 𝚜𝚝𝚛𝚒𝚙𝚎 𝚙𝚛𝚘𝚓𝚎𝚌𝚝𝚜 𝚊𝚍𝚍 𝚠𝚘𝚛𝚔𝚘𝚜/𝚊𝚞𝚝𝚑 https://t.co/KDYdGS8yg3 - **kathyyliao** (2026-04-29): https://x.com/kathyyliao/status/2049549320316485645 - you can't watch every browser session live. but when something goes wrong, you need to know what happened @CloudflareDev's Browser Run has Session Recordings. replay the full session, scrub the timeline, see exactly what your agent did here: a Browser Run session shopping Sentry's store for a bomber jacket - **NickVanWig** (2026-04-29): https://x.com/NickVanWig/status/2049548188034404756 - now available in another cli https://t.co/RcB3Aj9XBy - **collision** (2026-04-29): https://x.com/collision/status/2049547995625205797 - Increasingly, the best part of using Stripe is the millions of other companies using Stripe. How we use the network improve the product has been a big focus this year. Here are some of the networked ships from Stripe Sessions this week: Fraud. Radar is trained on signals from across Stripe, which now just sees most internet users and most payments. If a bad actor signs up for your product, we've generally already seen their device fingerprint, their email, or their card behavior—on someone else's business. For one AI company, 80% of the bad actors Radar caught had sailed right through their prior anti-fraud provider. As Stripe grows, the better every business on Stripe is protected. Link started as a way to save your payment details and has grown into a network of more than 250 million consumers. Link now stores stablecoins, powers agent wallets, and drives a 5% conversion lift for returning customers. Whenever a user signs up with Link on one business, every other Stripe business benefits the next time that customer checks out. Money movement. It turns out that Stripe businesses pay each other 4.8 million times a day. So we built instant, free transfers between Stripe Treasury accounts. Intelligence. 1.6% of global GDP now runs through Stripe; over 70 trillion data points last year. We've historically used that data to power our own products (Radar, authorization optimization). But now we’re putting it directly in your hands with Stripe Signals. Send us a customer, a transaction, a business—on or off Stripe—and we return a real-time risk score and explanation. Here's everything we announced this morning: https://t.co/UZWRHYTMrP. - **thearmadalabs** (2026-04-29): https://x.com/thearmadalabs/status/2049546677426561069 - Super proud of the team. Test it out and buy some domains! https://t.co/uNbVUJPUim - **SidChatterjee** (2026-04-29): https://x.com/SidChatterjee/status/2049545999673110915 - You can now create a new @CloudflareDev account and sign up for R2 and create a bucket, buy a domain and so much more ALL from the command line courtesy of @stripe Projects Have your agent try it today 😉 https://t.co/AQHlSbNWrr https://t.co/g5ebQ1is2F - **kathyyliao** (2026-04-29): https://x.com/kathyyliao/status/2049545241120895249 - broken sites are bad first impressions have your agent test user flows, catch broken links, + check performance before your users even notice a problem! https://t.co/1iKqE4y81M - **Cloudflare** (2026-04-29): https://x.com/Cloudflare/status/2049545195914498139 - Starting today, agents can now be Cloudflare customers. They can create a Cloudflare account, start a paid subscription, register a domain, and get back an API token to deploy code right away. https://t.co/qFgCivQTTi - **patrickc** (2026-04-29): https://x.com/patrickc/status/2049537446904483846 - @qibinlou @link Nope, any business - **OpenAI** (2026-04-29): https://x.com/OpenAI/status/2049534651702956103 - OpenAI DevDay is back. San Francisco September 29 - **thdxr** (2026-04-29): https://x.com/thdxr/status/2049532744934306064 - been a year since this post and we have definitely shifted towards liking more boilerplate-y explicit code vanilla used to mean simple but now it feels like a guarantee of ai going crazy and slopping shit up has anyone else felt this? https://t.co/o2oF2z83Sp - **mrramibanna** (2026-04-29): https://x.com/mrramibanna/status/2049531600107716636 - we did a thing. a really fun and agentic thing... https://t.co/2DG2dhNntL - **wolfejosh** (2026-04-29): https://x.com/wolfejosh/status/2049529783558570477 - 1/ I just finished one of the best books I've EVER read. Comes out next week. My friend @DavidEpstein (Range, The Sports Gene) has written a masterpiece on the virtue of constraints, of not thinking OUTSIDE the box but "INSIDE the BOX"... As a few quick excerpts that 🤯... https://t.co/VkjmzggHAv - **jeff_weinstein** (2026-04-29): https://x.com/jeff_weinstein/status/2049523328290025612 - livestream time. https://t.co/WZgiFqEAtr - **eve_bouff** (2026-04-29): https://x.com/eve_bouff/status/2049516552735863225 - You do your best work when you’re having lots of fun and don’t care about anything else other than having fun. Fun should always be the goal. - **_ashleypeacock** (2026-04-29): https://x.com/_ashleypeacock/status/2049512832895459437 - "What is a Durable Object?" Look no further than @aarondfrancis' video series, comprised of bite-sized videos covering the major features of Durable Objects Watch for free at https://t.co/y1D1yt6z2H https://t.co/vAah236CUo - **gustavscirulis** (2026-04-29): https://x.com/gustavscirulis/status/2049495063714029731 - How to turn any codebase into help articles your AI agent can use in ~15 minutes: 1. Install: npx skills add gustavscirulis/help-content-from-code 2. Run it → discovers how your product works and writes help content 3. Upload with Fin CLI 4. Test and go live Demo ↓ https://t.co/EerQ0cdYeO - **sama** (2026-04-29): https://x.com/sama/status/2049493609028923826 - feels like codex is having a chatgpt moment - **sama** (2026-04-29): https://x.com/sama/status/2049492646792757552 - i love that the team does stuff like this https://t.co/4tosXaZYGK - **whoiskatrin** (2026-04-29): https://x.com/whoiskatrin/status/2049449800701985277 - once i built an internal tool people actually use and now i haven’t tweeted in days - **RhysSullivan** (2026-04-29): https://x.com/RhysSullivan/status/2049363186286231606 - @thsottiaux @b_nnett this is such a better response than another company i can think of - **sama** (2026-04-29): https://x.com/sama/status/2049314532892475791 - https://t.co/WZA5vxK186 - **MattieTK** (2026-04-28): https://x.com/MattieTK/status/2049275929726234715 - In a month since launch EmDash has gone from... 0 stars to 10,000+, 0 downloads to 30,000+ a week, 1 contributor to 100+ contributors Thanks to everyone joining us on building a better CMS for the Internet. https://t.co/51meTXALAm - **garrytan** (2026-04-28): https://x.com/garrytan/status/2049222647867936871 - Deli Board is the best sandwich in SF and the world @bettersoma @war24182236 https://t.co/OouglpuqjH - **theallinpod** (2026-04-28): https://x.com/theallinpod/status/2049191973580329465 - David Friedberg: Is This Pesticide Causing a Colon Cancer Spike in Young People? Colon cancer rates in people aged 15-39 are up 80% over the last 20 years. A team from the University of Barcelona might have figured out why: Picloram exposure - **vimtor** (2026-04-28): https://x.com/vimtor/status/2049191474110026118 - sst now lets you easily add redirects and aliases to cloudflare components no need to manually define page rules or reach for the dashboard anymore https://t.co/1fbAeuh1QD - **kathyyliao** (2026-04-28): https://x.com/kathyyliao/status/2049186985261293578 - Browser Run lets you watch your agent's browser in real time and step in when it needs help here's my agent checking Google for up-to-date hotel pricing in Kyoto https://t.co/rhjPI5oWJz - **kimmaicutler** (2026-04-28): https://x.com/kimmaicutler/status/2049155285588303927 - Jerry Brown and Gavin Newsom, like them or not, were very predictable governors and steadiness and predictability matter for starting businesses and employing people, etc. https://t.co/Sc9c2P7MGG - **adamdangelo** (2026-04-28): https://x.com/adamdangelo/status/2049145483462148450 - I want to be able to tell my Waymo to take 280 instead of 101 - **ritakozlov** (2026-04-28): https://x.com/ritakozlov/status/2049114361524781419 - i've picked up the pen so many times to write about being a woman in tech and every time i chicken out because there's this catch-22: to talk about being a woman in tech, you need to have credibility. and once you start talking about it as a woman, you lose said credibility so i'm going to mortgage some of my credibility to get this off my chest, as someone who has both had a pretty successful career in tech, and leads a team with a lot of women on it: every woman you work with has had the most insane shit happen to her — on an almost daily basis. shit that makes you look at the camera and go "how did i end up here". from wild remarks about appearance to stalking and trauma dumping, and just constant dismissal from so many directions (employees, customers...). shit that you never tell anyone because they wouldn't believe you... i recently learned that like 97% of my followers on here are men. so my challenge to you is just to sit with that for a moment. you don't need to do anything about it (other than try not to be that person). but you should be aware that that's what every woman you work with deals with - **thsottiaux** (2026-04-28): https://x.com/thsottiaux/status/2048997818673537399 - Don't just reset Codex rate limits for fun, it costs money. Don't just reset Codex rate limits for fun, it costs money. ... but the vibes are good ... I have reset Codex rate limits for ALL paid plans to celebrate a good week and allow everyone to build more with GPT-5.5. Enjoy - **thdxr** (2026-04-28): https://x.com/thdxr/status/2048991814183473302 - deepseek is outrageously cheap even if you ignore the current 75% discount input token prices are 35x cheaper than opus cached tokens are 178x cheaper - **dok2001** (2026-04-28): https://x.com/dok2001/status/2048942247580696727 - @mitsuhiko @irvinebroque @southpolesteve need better answer here - **JoeFernandez** (2026-04-28): https://x.com/JoeFernandez/status/2048930762419007624 - One thing I see in my neighborhood in LA with Prop 13 is that there are at least 8 houses in a 2 block radius that have been empty for 5+ years. People who owned them have passed and it's so cheap for their kids just to leave it sitting. Bad for the neighborhood and city - **jonbeckman** (2026-04-28): https://x.com/jonbeckman/status/2048922230764036296 - Cloudflare DOs sync'd to the underlying Linear or Github Issue, using Codex or OpenCode in a sandbox, would be a cool implementation of this spec. https://t.co/oIov1ArRgz - **hiradp** (2026-04-28): https://x.com/hiradp/status/2048919026043396577 - migrated all my personal projects to @Cloudflare over the weekend including transferring all of my domains. big fan - **zeke** (2026-04-27): https://x.com/zeke/status/2048908566275453203 - OpenAI released their Privacy Filter model last week. It detects and redacts personally identifiable information in text: names, emails, phone numbers, addresses, API keys, and more. I packaged it with Cog and you can run it on Replicate: https://t.co/uF1tIot50L - **evanyou** (2026-04-27): https://x.com/evanyou/status/2048875087408120185 - Keep an eye on this https://t.co/7Z8nKEcQ9r - **RhysSullivan** (2026-04-27): https://x.com/RhysSullivan/status/2048874549069197692 - @badlogicgames Pi has been such a huge inspiration to what I’m working on There’s this incredible future of malleable software - **badlogicgames** (2026-04-27): https://x.com/badlogicgames/status/2048873906128216439 - @RhysSullivan you get me - **jeff_weinstein** (2026-04-27): https://x.com/jeff_weinstein/status/2048865799298076780 - Parallel (@p0) is a great case study for building for, and charging, agents. 1/ Agents can use Parallel api, cli, mcp, skill, etc to perform web search and tasks 2/ Parallel accepts pay-per-request machine payments over MPP and x402 (both protocols implemented via @stripe in a few lines of code): https://t.co/pvYsc43L8I 3/ Developers and agents can test machine payments using https://t.co/Xl3Vo3rIb7 4/ Agent-readable pricing https://t.co/conNmMGHVt 5/ Interactive agentic demo: https://t.co/zAyC4cGY1G - **RhysSullivan** (2026-04-27): https://x.com/RhysSullivan/status/2048851183595561416 - Dynamic Software will be a huge product You can see what it looks like today w/ this demo I made, it's PI level extensibility but able to be applied to any product and secure Give this to your coding agent and tell it to extend it with whatever you want https://t.co/JM6j7777GJ https://t.co/tZIdzDrXeL https://t.co/cHZv9tI4Le - **fin_ai** (2026-04-27): https://x.com/fin_ai/status/2048840572568531017 - Manual QA is broken. It’s slow, reactive, and only ever looks at a small slice of your conversations. That’s why we built Monitors, an always on QA system that evaluates every conversation across @Fin_ai and your human team. It scores quality against your standards, with human review layered in where it matters. Learn more about it below 👇 - **ycombinator** (2026-04-27): https://x.com/ycombinator/status/2048834287466868987 - AI for Low-Pesticide Agriculture @garrytan Farmers are stuck in a bad loop: use more chemicals, get diminishing results, pay more, take on more risk. And they can't just stop, because if pests win, crops die. AI that can identify individual weeds in real time, robotics that can treat one plant instead of blanketing a field, and new biological solutions mean this problem finally looks solvable. - **michellechen** (2026-04-27): https://x.com/michellechen/status/2048834092414775735 - 2 weeks out from agents week, long overdue message of appreciation to the ai platform team here at cf 🧡 there was a total of 10 (!!) agent week launches that were built on products like workers ai and ai gateway. we've become an integral part of the @CloudflareDev platform - i'm proud of the team and nostalgic that we've come such a long way - **Cameron_C2** (2026-04-27): https://x.com/Cameron_C2/status/2048820452454260857 - I'm working on a RFC for embedded framework support in oxlint. This aims to add full support for languages such as Vue, Astro & Svelte, with type aware linting, type checking! Keep an eye out later this week for the RFC. - **dwr** (2026-04-27): https://x.com/dwr/status/2048764030165057829 - Who’s doing the most interesting research on how to combat invasive insects and pathogens destroying North American trees? Citrus greening, emerald ash borer, Dutch elm, hemlock woolly adelgid, chestnut blight, etc. - **levelsio** (2026-04-27): https://x.com/levelsio/status/2048703164807868455 - llms.txt did indeed turn out to be complete bullshit https://t.co/pRn8n2w8fV https://t.co/PCKi1fS0lj - **fin_ai** (2026-04-27): https://x.com/fin_ai/status/2048679878854066580 - Support doesn’t just happen in tickets anymore. It happens in communities. So we’ve launched @Fin_ai over @Discord, built for how support and conversation coexist in this channel. Discord is different. Some questions belong in public. Others need to go private, fast. Fin handles both, and you control when and how it steps in. → 𝗬𝗼𝘂 𝗰𝗼𝗻𝘁𝗿𝗼𝗹 𝗲𝗻𝗴𝗮𝗴𝗲𝗺𝗲𝗻𝘁: Trigger responses through mentions, emoji reactions, or keywords. → 𝗣𝘂𝗯𝗹𝗶𝗰 𝗮𝗻𝗱 𝗽𝗿𝗶𝘃𝗮𝘁𝗲, 𝗱𝗼𝗻𝗲 𝗿𝗶𝗴𝗵𝘁: Answer in-thread, then move sensitive issues into private conversations. → 𝗕𝘂𝗶𝗹𝘁 𝗳𝗼𝗿 𝗗𝗶𝘀𝗰𝗼𝗿𝗱: Community and support, side by side, with the right boundaries. Brian walks through how it works below 👇 - **boristane** (2026-04-27): https://x.com/boristane/status/2048579608665374792 - coming home to see that https://t.co/vSIX2IxOgK investigated and fixed an issue while I was out has got to be top feeling - **sama** (2026-04-26): https://x.com/sama/status/2048552677433643427 - Our Principles: Democratization, Empowerment, Universal Prosperity, Resilience, and Adaptability https://t.co/g58WGndARw - **badlogicgames** (2026-04-26): https://x.com/badlogicgames/status/2048516965682368846 - It would be lovely if inference engines would have endpoints that advertise: - all locally cached models and their specs (input modalities, context window size, thinking, tool caling) - loaded model(s) that would allow harnesses to easily enumerate things dynamically, instead of having users (or their agents) write silly config files. - **MVP_Mindset** (2026-04-26): https://x.com/MVP_Mindset/status/2048508670330495051 - Steph Curry shares the smallest detail in his game that helps separate him from others. "Breathing is a skill…Controlling your breathing, recovery, controlling your nerves…I’ve really tried to master that." There’s a reason the breath is often called the bridge between the mind and body. And the best part? You always have access to it. No matter the moment. No matter the pressure. No matter the environment. It’s your fastest way to recover and your most reliable way to slow the game down and reset. 📹: SLAM NBA - **theCTO** (2026-04-26): https://x.com/theCTO/status/2048504395063513535 - The age of serverless email infrastructure https://t.co/zUbZ52zqyG - **boristane** (2026-04-26): https://x.com/boristane/status/2048446003783602425 - building self-operating software https://t.co/5R13oE0Eft - **nurodev** (2026-04-26): https://x.com/nurodev/status/2048442870718226607 - @kisyphut @vikingmute This, and a few others, is next on our list to add soon - **sama** (2026-04-26): https://x.com/sama/status/2048428561481265539 - feels like a good time to seriously rethink how operating systems and user interfaces are designed (also the internet; there should be a protocol that is equally usable by people and agents) - **vikingmute** (2026-04-26): https://x.com/vikingmute/status/2048404675960139788 - Cloudflare 新版 wrangler 的 Local Explorer 真好用，总算有个 web 界面可以看本地的 CF 资源了，升级到最新版，运行 wrangler dev 命令以后，在 console 中再按 e，就会弹出一个 localhost:8787/cdn-cgi/explorer 这样的界面，就可以看本地所有的 CF 资源了，比原来方便好多。 https://t.co/DbrHIrUczQ - **pmarca** (2026-04-26): https://x.com/pmarca/status/2048221762991910974 - When something becomes abundant and cheap, someone else becomes scarce and valuable. https://t.co/tsKkuyq4ML - **kimmaicutler** (2026-04-25): https://x.com/kimmaicutler/status/2048129008194433194 - There’s only one candidate in the SF Congressional race with the receipts to show they’ve been working on housing for 10+ years. There are finally towers being proposed near me in the last month. Would have been impossible without him. https://t.co/YD6zPiJGzD - **mattzcarey** (2026-04-25): https://x.com/mattzcarey/status/2048102657512935626 - My talk at AI Engineer “Every API is a Tool for Agents” is out on YouTube https://t.co/hk1aeZ9IsH Thanks to @swyx and the @aiDotEngineer crew for having me :) - **Teknium** (2026-04-25): https://x.com/Teknium/status/2047869295686975529 - I literally run 12 hermes agent instances every day in parallel to build Hermes Agent, and its now a top 100 GitHub repositories of all time. Agents do bring value and do create substantive software and work. https://t.co/sIjX6ooAca - **sama** (2026-04-24): https://x.com/sama/status/2047823357635354814 - this was a good week. proud of the team. happy building! - **tobi** (2026-04-24): https://x.com/tobi/status/2047816105163444653 - @badlogicgames PI is now over 50% of agent usage at Shopify internally. - **badlogicgames** (2026-04-24): https://x.com/badlogicgames/status/2047802729947426883 - i'm sorry, @threepointone made me do it... https://t.co/RDUp1Oal0j - **zeeg** (2026-04-24): https://x.com/zeeg/status/2047797882758070630 - Everyone is slowly coming to this realization, and I assure you, no one is running multitudes of agents overnight. No one that is doing anything of substance at least. There _are_ people pretending to be scientists, or fully caught up in their drug infused AI overdose, that think their slop machines are changing the world. They're not tho, and they're just wasting a bunch of money and compute to create a lot of LoC that will just get thrown away. The state of the art is still "can we even one shot a production quality patch that we wont regret later", and its rarer than you'd expect based on discourse. - **pingToven** (2026-04-24): https://x.com/pingToven/status/2047726369778385046 - @badlogicgames @pvncher @deepseek_ai 🙃 i am specifically trying to validate rn that all the third party inference providers handle this properly too lol - **thdxr** (2026-04-24): https://x.com/thdxr/status/2047669263633178672 - deepseek v4 can run on huawei chips another step towards self reliance thanks to our own policies can't believe how many times we'll make this mistake - **evanyou** (2026-04-24): https://x.com/evanyou/status/2047644304525623389 - Both Codex and Claude Code’s desktop apps are using Oxlint & Oxfmt :) Codex also already on Vite 8! - **cramforce** (2026-04-24): https://x.com/cramforce/status/2047505675719512507 - If @mikeal was still alive he'd tell you that the best fried chicken sando both west and east of Japan is at Spinning Bones in Alameda. What a master piece 😿 - **lessin** (2026-04-23): https://x.com/lessin/status/2047444069996249464 - This is hilarious and smart... I am actually sad I didn't think of it! https://t.co/F3nLP754rW - **KentonVarda** (2026-04-23): https://x.com/KentonVarda/status/2047428696294789294 - Nitpick: It's not quite correct to say that using an old compatibility date means using an old "version" of the runtime. When hosted on Cloudflare, you are always using the latest version, but the flags tell it to emulate old versions in certain respects. Note that API changes that aren't breaking aren't gated by flags, so even if you have a 2019 compatibility date on your Worker, it will still be able to see lots of new APIs that were introduced since then -- just not the (usually minor) breaking changes. - **adityaag** (2026-04-23): https://x.com/adityaag/status/2047421448449630379 - How does one make sense of SF? We are living through the craziest computing revolution of the last 2 decades....and 95% of the value being created is within a 4-mile radius in SF (h/t @eladgil ) It's just hard to understand....why is the concentration so high? What is it about this place? There are good reasons that people will offer...good universities, great companies, talent networks, venture capital yada yada yada But all of that is hard to emotionally understand. I tend to try to answer these questions through my own lived experience. Two things happened this week that brought it home for me. On Wed evening, two GOATS of the design world (@soleio and @rsms ) did a design talk at SPC. I saw some other GOATS at the talk (@aaron_ , @benbarry ) And you realize that these GOATS are just so curious about the world. They just want to build and learn. And give back. And with an air of humility and passion. It's so so cool and inspiring. And then this week we had someone release a mind-bendingly amazing demo of every pixel being generated. https://t.co/CBBqdEaUaS People were blown away but @eddiejiao_obj and @zan2434 had been jamming creatively for a bunch of time @spc . And in that entire time, they were surrounded by a culture and environment and ecosystem that says "that is crazy but my god, if it works, then what could happen" There is wonder, amazement and curiosity. And passion. I wish this city were cleaner and better governed (but Dan Lurie is crushing). We are so lucky to be surrounded by weirdos who love to build. This is why we are who we are. - **syeefk** (2026-04-23): https://x.com/syeefk/status/2047411372518998387 - @BalintFerenczy @irvinebroque This is beautiful. Congratulations! - **venturetwins** (2026-04-23): https://x.com/venturetwins/status/2047409327481507937 - Crazy how you can just...ship a model without a giant PR campaign to scare the shit out of everyone first - **sama** (2026-04-23): https://x.com/sama/status/2047403771416940715 - "don't retweet this, don't retweet this, don't retweet this..." ah fuck it, life imitates art. https://t.co/5HYvYXfQzd - **sama** (2026-04-23): https://x.com/sama/status/2047379615589777666 - 1. We believe in iterative deployment; although GPT-5.5 is already a smart model, we expect rapid improvements. Iterative deployment is a big part of our safety strategy; we believe the world will be best equipped to win at the team sport of AI resilience this way. 2. We believe in democratization. We want people to be able to use lots of AI; we aim to have the most efficient models, the most efficient inference stack, and the most compute. We want our users to have access to the best technology and for everyone to have equal opportunity. We have been tracking cybersecurity as a preparedness category for a long time, and have built mitigations we believe in that enable us to make capable models broadly available. 3. We love you and we want you to win. We want to be a platform for every company, scientist, entrepreneur, and person. (My whole career has largely been about the magic of startups, and I think we are about to see that magic at hyperscale.) - **OpenAI** (2026-04-23): https://x.com/OpenAI/status/2047376561205325845 - Introducing GPT-5.5 A new class of intelligence for real work and powering agents, built to understand complex goals, use tools, check its work, and carry more tasks through to completion. It marks a new way of getting computer work done. Now available in ChatGPT and Codex. https://t.co/rPLTk99ZH5 - **jordwalke** (2026-04-23): https://x.com/jordwalke/status/2047375747409424846 - Introducing real iOS and Android device testing, right from your browser. If you've ever: 👉built a @Replit mobile app and had no way to test it on a phone 👉 been solely on Windows/Android and don't touch iOS …this is for you! https://t.co/Rr27zJUAy9 - **kathyyliao** (2026-04-23): https://x.com/kathyyliao/status/2047375039436341301 - called out by @davidliu514 for scoring 0 on https://t.co/hFqTU6lo4K so I went from 0 → 100 in one weekend added: robots.txt, sitemap, llms.txt, MCP WebMCP, markdown endpoints, +more https://t.co/RzZbUw6uLZ https://t.co/an29I6Ax6B - **jeff_weinstein** (2026-04-23): https://x.com/jeff_weinstein/status/2047368670113062979 - props to @cloudflare for building https://t.co/XL3ydESIqb, which delightfully does what it says. - **rsproule** (2026-04-23): https://x.com/rsproule/status/2047358538469540032 - @irvinebroque We had an early version where there was a TXT record on the DNS that would point to where the discovery document lived in cases where server couldn't do /openapi.json. We are very open to supporting more locations - **dwr** (2026-04-23): https://x.com/dwr/status/2047355077736706551 - Cool to see Cloudflare working with @mpp! https://t.co/yLI3pLFeAE - **sama** (2026-04-23): https://x.com/sama/status/2047347674743963705 - we love seeing our users win. we want to give you the best tools, lots of compute, and watch you do the magic. - **thsottiaux** (2026-04-23): https://x.com/thsottiaux/status/2047322142262641099 - @badlogicgames Not true. Please don't spread misinformation out there, it's already hard enough to keep facts straight! Will investigate today if there are differences but the requests should all go to the same clusters irrespective of the client (pi, opencode, openclaw, w/e). - **BalintFerenczy** (2026-04-23): https://x.com/BalintFerenczy/status/2047318552898642319 - I joined Cloudflare as a design engineer! Made some clouds. https://t.co/r4ePJBjf7F - **celso** (2026-04-23): https://x.com/celso/status/2047254802061541447 - It never fails as an ice breaker. https://t.co/bjNLDrRSfX - **mdo** (2026-04-23): https://x.com/mdo/status/2047182412908794129 - Thanks everyone for the love with the launch of https://t.co/nr5toR1e3V today! It was a ton of fun to design and build this alongside @SlexAxton and the rest of @pierrecomputer. Excited to keep improving Trees and Diffs, so let us know what you think! https://t.co/c8mSL0LKei - **thearmadalabs** (2026-04-23): https://x.com/thearmadalabs/status/2047149036222955868 - This kind of blew my mind. Also great domain 🤯 https://t.co/GpQhvbQyAN - **fidjissimo** (2026-04-23): https://x.com/fidjissimo/status/2047131244618744150 - So excited for these launches and grateful for the team working hard on this. https://t.co/4zaU1iXy5Y - **sinasanm** (2026-04-23): https://x.com/sinasanm/status/2047130501098749964 - blows my mind that i was able to build a claude code alternative with kimi k2.6 on cloudflare, get to feature parity over night to the point that i now use it instead of claude code. fast and cheap and same quality - **s8mb** (2026-04-23): https://x.com/s8mb/status/2047125440847106121 - More airports should follow SFO and ban loudspeaker announcements. Makes a big difference to how tiring it is to be there. - **aglev33** (2026-04-23): https://x.com/aglev33/status/2047105006663606325 - Preliminary application to replace the Trader Joe's steps from the Rockridge BART station with a pair of 31-story and 25-story towers with a total of 415 market-rate senior living units: https://t.co/2ulqsr2JMn. https://t.co/zFqpeERhE3 - **nateberkopec** (2026-04-22): https://x.com/nateberkopec/status/2047057328541979130 - Scanlan is a legend, this is required viewing. https://t.co/xB2qKgtj9w - **badlogicgames** (2026-04-22): https://x.com/badlogicgames/status/2047038534264013258 - People of pi. Cloudflare release with a minor breaking change. - JSON validation no longer relies on AJV, but uses Typebox. Update your existing extensions as per changelog (@sinclair/typebox -> @typebox) - OSC 9;4 indicator in supporting terminals for @lucasmeijer thanks to MVP @cinnamon_msft - Tools can now signal "terminate the loop" so you can do structured output and other shenanigans easily - Custom auto completion providers via extension. This one fetches the last 100 isues from GH when i type # and let's me search through them via type ahead And lots of bug fixes and internal refactoring. - **brian_scanlan** (2026-04-22): https://x.com/brian_scanlan/status/2047029394980413658 - It has been DAYS since my last podcast appearance, so here’s another one. This is a particularly fun chat with the incredible @robbyrussell on the On Rails podcast. https://t.co/OCGSS5QrKd - **jonas** (2026-04-22): https://x.com/jonas/status/2047009234085114063 - @irvinebroque For platform use cases I would love to see 1) arbitrarily tag resources with customer ids and other grouping tags (kinda like you can do for LLM requests in AI gateway) 2) breakdown of worker costs by DO namespace and ID - **ciaran_lee** (2026-04-22): https://x.com/ciaran_lee/status/2046987288983261638 - I think this is pretty cool: https://t.co/T7LpI43YZL The timeline is accelerating, more new stuff will be coming soon :) https://t.co/ThM7iiGQHG - **guybedford** (2026-04-22): https://x.com/guybedford/status/2046983298099093601 - We recently undertook an effort to improve reliability of Rust Workers, which involved implementing panic=unwind and abort recovery in wasm-bindgen using WebAssembly Exception Handling. Excited to be able to share this work! https://t.co/urPwwPY2Vm - **jonas** (2026-04-22): https://x.com/jonas/status/2046949972239638661 - Full stack tanstack start app bundled into a dynamic worker durable object facet has been achieved internally - **Cloudflare** (2026-04-22): https://x.com/Cloudflare/status/2046938329594155077 - Panics in Rust Workers were historically fatal, poisoning the entire instance. By collaborating upstream on the wasm‑bindgen project, Rust Workers now support resilient critical error recovery, including panic unwinding using WebAssembly Exception Handling. https://t.co/pRx24SWg7L - **TheRealAdamG** (2026-04-22): https://x.com/TheRealAdamG/status/2046918440632320442 - You forgot to tag me. :) To quote Billy Joel “We didn’t start the fire”. I, personally, haven’t forgotten about those petty Super Bowl ads. Those ads that mocked OpenAI for serving (high quality, no intrusive) ads to make sure we can continue to have a robust free and $8 tier offering. My snark may have too much bite at times, and for that I apologize, but if that’s the reason you want Anthrophic to “win”.. sigh. - **zeke** (2026-04-22): https://x.com/zeke/status/2046814555956011510 - The prompting skills are intentionally model-agnostic. The hottest image model today might be old news by next week, so instead of being prescriptive about specific models, they teach general-purpose techniques: prompt-images: photographic language, text rendering, style transfer, character consistency, inpainting prompt-videos: scene description, camera motion, audio and dialogue, time-coded multi-shot prompting - **zeke** (2026-04-22): https://x.com/zeke/status/2046814544035811473 - I just added two new official Replicate skills for prompting image and video models. This little skillset is all your agent needs to find, compare, prompt, and run all the best AI models on Replicate. https://t.co/qXGRfdQk6P - **zeke** (2026-04-22): https://x.com/zeke/status/2046812115793760566 - I made an agent skill to speed up Chrome DevTools MCP. npx skills add zeke/faster-chrome-devtools-skill https://t.co/ivQgm13p1c https://t.co/nOPp7RDdnn - **sama** (2026-04-22): https://x.com/sama/status/2046808114561974567 - @TheAmolAvasare ok boomer - **ritakozlov** (2026-04-22): https://x.com/ritakozlov/status/2046747108670136342 - family dinner featuring @threepointone’s famous curry with the best people @dillon_mulroy @michellechen @dinasaur_404 @fatimayusf https://t.co/wZmLP2TVnq - **jeff_weinstein** (2026-04-21): https://x.com/jeff_weinstein/status/2046738459336020417 - machine payments for nyc live transit data https://t.co/eiiPfG8d0R - **jeff_weinstein** (2026-04-21): https://x.com/jeff_weinstein/status/2046682966471172505 - > User: ~"Tested it. Quite seamless. I'm impressed. Experience was really well done. Will likely ship to production." > Me: "If you had to give a single piece of critical feedback or suggestion (or two), what would it be?" > User: "Well... [INSANELY DEEP COMMENTS & QUESTIONS]" - **YIMBYLAND** (2026-04-21): https://x.com/YIMBYLAND/status/2046670669912391780 - What the fuck are we even doing here? https://t.co/Kn4aC1QHw1 - **jonas** (2026-04-21): https://x.com/jonas/status/2046663704775725241 - I made this extremely rough and ready cloudflare artifact viewer The entirely vibe-slopped code is here https://t.co/j1Rywe9Ejp An example deployment is here: https://t.co/ntL4TET7bv If anyone is interested, I'll polish it up a bit tomorrow https://t.co/xoe4jpvlNj - **vimtor** (2026-04-21): https://x.com/vimtor/status/2046650374551916894 - sst now allows you to deploy cloudflare hyperdrive the docs include examples for @planetscale and zero trust tunnel for aws postgres https://t.co/OYymyBpuyM - **LakeAustinBlvd** (2026-04-21): https://x.com/LakeAustinBlvd/status/2046641235826229653 - About to hit the eight year mark at Cloudflare and I have never been more excited about what is being built here. Every day now is just: DevPlat team showcases something new & amazing -> we bring it into internal tools a couple hours later -> feedback makes it better for customers -> repeat. - **brian_scanlan** (2026-04-21): https://x.com/brian_scanlan/status/2046634306710233099 - There’s a lot more detail in this blog post about how we’ve implemented automatic pull request approvals. We think they are better than human code review for the vast majority of changes. Faster, more reliable, safer - choose all three. https://t.co/eQUlty603l - **grinich** (2026-04-21): https://x.com/grinich/status/2046618013160587676 - good advice from @boztank https://t.co/fLZ6Y3qGFt - **vimtor** (2026-04-21): https://x.com/vimtor/status/2046568046714576910 - now if you create a new sst resource inside a cloudflare monorepo, only the root 𝚜𝚜𝚝-𝚎𝚗𝚟.𝚍.𝚝𝚜 will change we've also added a config to let you exclude directories from type generation for special cases https://t.co/qUQpWBJb0E - **samlambert** (2026-04-21): https://x.com/samlambert/status/2046383280795320468 - Bare metal with elasticity https://t.co/W2fCwnoQ1G - **nxthompson** (2026-04-21): https://x.com/nxthompson/status/2046380148493005174 - The stats here are kind of remarkable. BART's new fare gates have led to a 1,000-hour decline in clean up time; 41% drop in crime; and $10 million increase in projected revenue. https://t.co/bQMIcgUPal https://t.co/SkKqcW7Cp2 - **taylorotwell** (2026-04-20): https://x.com/taylorotwell/status/2046368273038553121 - Job debouncing and Cloudflare email support in tomorrow’s Laravel release. ✨ - **akaphill** (2026-04-20): https://x.com/akaphill/status/2046361015693320464 - You can now query Cloudflare logs from R2 SQL, DuckDB, Snowflake, or anything else that reads Iceberg. https://t.co/yKL35cecVg https://t.co/yC99a8sdPA - **dillon_mulroy** (2026-04-20): https://x.com/dillon_mulroy/status/2046322228011372923 - artifacts released on thursday and in production and a conference talk by monday. so cool @kentcdodds https://t.co/sRrJfGDBkg - **Noahpinion** (2026-04-20): https://x.com/Noahpinion/status/2046322075644907636 - Property taxes are better than income taxes because property can't move https://t.co/yB5hSkDxIx - **michellechen** (2026-04-20): https://x.com/michellechen/status/2046302713865175497 - kimi k2.6 on @CloudflareDev workers ai, now live on @OpenRouter 🌙 https://t.co/CKYqIndT90 - **michellechen** (2026-04-20): https://x.com/michellechen/status/2046297037742997909 - partnering with @Kimi_Moonshot to bring kimi k2.6 to @CloudflareDev workers ai on day 0 better for coding and agentic use cases! try it out now: https://t.co/TYn3uZDRY5 - **clairevo** (2026-04-20): https://x.com/clairevo/status/2046292426793582913 - I have been hyping @brian_scanlan and @intercom to every eng leader I've spoken to since getting a behind the scenes look at how they 2x'd their productivity in their R&D org. In this week's episode of How I AI, Brian pulls up dashboards and tools and shows us how it's done: - explicit targets + reporting on leading indicators of AI adoption - how they measured how quality increased with more AI usage - centralized skills repo for the entire company - telemetry on claude code + evals on internal success - building for an agent-first onboarding flow Sorry to be that bro, but truly: STOP what you are doing and watch this right now if you are looking to inflect your eng + product team with AI. As always, a huge thanks to our sponsors: 🧠Celigo - Intelligent automation built for AI: https://t.co/wXLyXeR53y @cursor_ai - The best (and my fave!) way to code with AI: https://t.co/GBqDSfjHdu On YT now 👉 https://t.co/YLItD2lKFc - **aninibread** (2026-04-20): https://x.com/aninibread/status/2046289191252939156 - Agents Week at @Cloudflare is a wrap. Here is a recap of everything we shipped this week. 🚢 https://t.co/YkYQNlYVSI - **danshipper** (2026-04-20): https://x.com/danshipper/status/2046280089818828835 - counterintuitive things I believe about AI: SaaS is not dead, it will be stronger than ever The userbase for SaaS will 10x over the next 3 years because agents will become users Knowledge work will change dramatically, but everyone will still have jobs (except very specific categories like personal injury insurance for car crashes) Humans will continue to be the sandwich at the beginning and end of every AI process (h/t @kieranklaassen and @trevin who came up with this) Personal software looks like OpenClaw, not a vibe-coded Salesforce clone You'll use Codex and Claude Cowork through your SaaS software, and use your SaaS software through Codex and Claude Code Because you'll use Codex with your SaaS, all software users will start to look like extremely technical users All SaaS will need to incorporate the idea of 'presence' for agents—making what agents are doing legible in real-time to users Agents are currently built for 1-1 interactions, they'll need to be built for one to many and many to many interactions Specialization will be just as important in agents as it is in humans, therefore we will live in a many-agent world Most software companies are not building for a world where everyone has an agent, and they're wasting a ton of time and capital developing capabilities that a user with Claude Code doesn't need - **RhysSullivan** (2026-04-20): https://x.com/RhysSullivan/status/2046278830743302286 - RLS and its consequences have been a disaster for the vibe coders https://t.co/r6R0y5wM9I - **brian_scanlan** (2026-04-20): https://x.com/brian_scanlan/status/2046263407146381333 - I was on the amazing How I AI podcast with @clairevo this week! I go through how we've been using Claude Code to double the throughput of our R&D org. (link follows...) - **zeeg** (2026-04-20): https://x.com/zeeg/status/2046251663909171489 - RLS was a mistake and folks exposing that level of complexity to less technical users is asking for trouble. It was a mistake in Firebase. It’s a mistake in Supabase. It will be a mistake in the next product too. I personally - even knowing how to secure it - would never touch it. It’s the worst security footgun you can imagine. One small mistake and your data is available to the world. - **ihteshamali** (2026-04-20): https://x.com/ihteshamali/status/2046244094255734903 - Cloudflare just open-sourced an email client where an AI agent reads your inbox, drafts your replies, and never sends anything without your permission. It's called Agentic Inbox. It runs entirely on Cloudflare Workers. Zero third-party servers touching your emails. Here's what the AI agent can actually do: → Reads your inbox and surfaces what matters → Searches across all your conversations → Drafts replies automatically the moment a new email lands → Holds every draft until you explicitly confirm it never sends on its own → 9 built-in email tools running on Workers AI Each mailbox lives in its own Durable Object with a SQLite database. Attachments go to R2. Your email never leaves your Cloudflare account. One click deploys the whole thing. Cloudflare provisions R2, Durable Objects, and Workers AI automatically. Apache 2.0. 100% Opensource https://t.co/VTOl8LVWmx - **vimtor** (2026-04-20): https://x.com/vimtor/status/2046239069420577234 - sst now lets you deploy @reactrouter on cloudflare this includes custom domains, caching, and much more https://t.co/Ow6JnNsJjQ - **rajesh_bh** (2026-04-20): https://x.com/rajesh_bh/status/2046234804069560325 - Very proud of this great work by @_ryanskidmore https://t.co/5i0G1EP1Kv - **elithrar** (2026-04-20): https://x.com/elithrar/status/2046227539790643599 - the team did an incredible job with this: we run automated review on every GitLab MR (with human oversight & override). we mix Kimi K2.5 + GLM 5.1 on Workers AI and Opus + GPT 5.4 based on MR size & complexity. https://t.co/ZI5saOtGHt - **_ryanskidmore** (2026-04-20): https://x.com/_ryanskidmore/status/2046223113990332572 - Ever wondered how a company like @Cloudflare handles AI Code review at scale? My very first post on the Cloudflare blog dives deep into exactly that! https://t.co/XzyfrhveoQ - **Cloudflare** (2026-04-20): https://x.com/Cloudflare/status/2046219878349050095 - We built our internal AI engineering stack on the same products we ship. That means 20 million requests routed through AI Gateway, 241 billion tokens processed, and inference running on Workers AI, serving more than 3,683 internal users. https://t.co/MyodrT9Pwg - **rajesh_bh** (2026-04-20): https://x.com/rajesh_bh/status/2046214853761888742 - How we built Cloudflare's internal AI engineering stack. 93% R&D adoption in 11 months 241B tokens/mo through AI Gateway Every MR gets an AI code review Built on the products we ship to customers Overview + architecture: https://t.co/M4gCqQT398 links to AI code review deep dive - **ivanleomk** (2026-04-20): https://x.com/ivanleomk/status/2046039879033946596 - TIL that you can deploy zig binaries on cloudflare - **mattrothenberg** (2026-04-19): https://x.com/mattrothenberg/status/2045670375191613871 - Can't believe I got nerd-sniped into rebuilding the Cloudflare Blog with Emdash. should we try to get this shipped soon? 👀 https://t.co/gxFjxMUb1R - **RhysSullivan** (2026-04-19): https://x.com/RhysSullivan/status/2045657774051533282 - man, it takes a bit of work but when you get a nice OTEL set up and your agent connected to it it's magical https://t.co/zFxjVg5uEx - **kane** (2026-04-19): https://x.com/kane/status/2045657668975796613 - At this point the board of supervisors @sfbos has held multiple hearings about Waymo while human drivers continue to vaporize pedestrians with impunity. https://t.co/wfucz6Qkrs https://t.co/NJFGgmQiBD - **vimtor** (2026-04-18): https://x.com/vimtor/status/2045590617627984266 - now you can deploy @cloudflare static sites using worker assets in sst this means you’ll get better defaults, faster deployments, redirects, and much more https://t.co/gKSLn73Its - **thdxr** (2026-04-18): https://x.com/thdxr/status/2045525102268162459 - are there people out there who just want to refactor every day? just wake up and find the worst code and just chip away at it and clean it up wake up the next day do it again, infinitely improving things with zero external impact? - **dillon_mulroy** (2026-04-18): https://x.com/dillon_mulroy/status/2045499301938315430 - @appfactory in my experience you just need to join random meetings and interject with “is this idempotent” and then leave - **sriramk** (2026-04-18): https://x.com/sriramk/status/2045434795329458201 - passkeys don't get enough credit for how much they have shaped and secured the internet these past 2-3 years. when a site has it, it is usually a positive signal on how seriously they take security. going from a world of 2fa auth codes that dont always sync / sites that only allowed SMS to this has been an undeniable upgrade. - **BenStiller** (2026-04-18): https://x.com/BenStiller/status/2045370082520445282 - I am a big Steve Kerr fan. - **thsottiaux** (2026-04-18): https://x.com/thsottiaux/status/2045299702590259631 - We are barely getting started with Codex. Good engineering and clear product thinking has never been more important and I’m proud of the team for deeply thinking about the primitives that will stand the test of time. Things will continue to accelerate from here. - **parkerconrad** (2026-04-17): https://x.com/parkerconrad/status/2045283443840532596 - Rory's takes are always very sharp. https://t.co/oSQdn8YBgX - **retlehs** (2026-04-17): https://x.com/retlehs/status/2045264249501499849 - ✅ https://t.co/d4fNgRAPYu https://t.co/H6YCNKe0BX - **mattzcarey** (2026-04-17): https://x.com/mattzcarey/status/2045256038581260787 - uno reverse first commit to the lib (gitty) which became our zig engine for Artifacts was 23rd March.... 2 days before yours... I've been working on this in one form or another since december. Starting with my zagi project, git cli in zig, sounds alot like ziggit ?? gitty was even open access for a few days but made private later that week. Please don't accuse people of things they didnt do - **dok2001** (2026-04-17): https://x.com/dok2001/status/2045252923479978457 - Lossless compression cut LLM weights by up to 22%. No quality loss. @Cloudflare Research Cloudflare Research built Unweight to attack the real bottleneck on H100s: memory bandwidth, not compute. Weights decompress straight into shared memory and feed tensor cores without the round trip. Blog: https://t.co/Vnxg0hvzWK Paper: https://t.co/JJ4WkLIER9 - **kevinweil** (2026-04-17): https://x.com/kevinweil/status/2045230426210648348 - Today is my last day at OpenAI, as OpenAI for Science is being decentralized into other research teams. It’s been a mind-expanding two years, from Chief Product Officer to joining the research team and starting OpenAI for Science. Accelerating science will be one of the most stunningly positive outcomes of our push to AGI, and I’m rooting for @sama @markchen90 @fidjissimo @gdb @merettm and the whole team! - **lessin** (2026-04-17): https://x.com/lessin/status/2045208254511526034 - Chat sucks.. the right interface for AI? my answer is email... I setup a bridge where i can email in to a private domain, ask for anything with an instance that has key data and spin up workers to process as many tasks in parallel as i want / reply when done.. https://t.co/y9hEvAG9Y0 - **threepointone** (2026-04-17): https://x.com/threepointone/status/2045197038883598720 - https://t.co/7N6WuNVcgf https://t.co/TZzlPfUKgZ - **kathyyliao** (2026-04-17): https://x.com/kathyyliao/status/2045195177720172554 - Agents will outnumber humans on the Internet, so sites need to be agent-ready --> WebMCP is one of the newest ways to do it. Launched by the @googlechrome team, it lets your site expose tools that agents can discover + call directly to test your site's WebMCP setup, @CloudflareDev Browser Run has an experimental pool of browsers running Chrome beta so you can verify your tools work before agents show up - **KentonVarda** (2026-04-17): https://x.com/KentonVarda/status/2045195153044828481 - It's kind of crazy how much of the way we've been designing Workers over the past 9 years unexpectedly turns out to be so relevant to AI and agents. Durable Objects and lightweight isolate sandboxes are obvious big things. But there are subtler things. Consider "bindings". In Workers, our environment (`env` object) doesn't just contain strings. It can contain live objects, which we often call "bindings". For instance, a Workers KV binding is a live object representing a Workers KV storage namespace. Once you've configured it, you can just do: let val = await env.MYKV.get("foo") await env.MYKV.put("foo", "new value"); Notice: There's no connection string. No secret token that you have to pass to talk to your KV namespace. The Workers Runtime handles it for you. You just get an already-initialized client object, on which you can call methods. You can still do everything you want to do. But you know what you can't do? Leak the secret token. Because there isn't one. A KV namespace binding fundamentally cannot be "leaked" because it's not bytes. But over the years, a lot of people have questioned whether this really mattered. I've had people inside and outside the team say: "Why are you so weird, Kenton? Yeah sure it can't leak but now I have to learn this new way of thinking about things. No other runtime works this way so writing portable code takes extra work. I'd rather just stick to what I'm used to, and anyway I know better than to leak my environment variables." Well, now we have AI agents writing the code and... suddenly everyone is worried about agents leaking keys. People are creating convoluted schemes to intercept the outbound traffic and inject keys in a proxy, or trying to issue very-short-lived keys so that if the agent leaks them the window of attack is short. Ahem. Welcome, folks! We solved this 8 years ago! Here's an old blog post -- written when I personally was still very much Not Thinking About AI -- which seems so much more relevant now: https://t.co/hHfMQpu4FH - **ackriv** (2026-04-17): https://x.com/ackriv/status/2045177696506794336 - Gzip: 92KB Shared dictionary: 159 bytes 99.5% smaller 🤯 Compression just got a memory. Blog + live demo: https://t.co/Zx23KmZUZj - **eastdakota** (2026-04-17): https://x.com/eastdakota/status/2045176827837120765 - Given how much more Internet traffic AI Agents are going to create, we need new strategies to ensure the network can keep up. Here’s one… https://t.co/I6v27ETAID - **KankaniAbhishek** (2026-04-17): https://x.com/KankaniAbhishek/status/2045174227251425635 - AI is writing a rapidly growing share of production code. Soon, agents will deploy it themselves: write, ship, observe, ramp, roll back, all without a human in the loop at every step. That only works if there's a safety primitive the agent can reason about. Feature flags are that primitive! - **KankaniAbhishek** (2026-04-17): https://x.com/KankaniAbhishek/status/2045169168237076748 - **Cloudflare has released Flagship thanks for the sneak peek @wongmjane 👀 https://t.co/31tvWC8XeX - **celso** (2026-04-17): https://x.com/celso/status/2045168454542503962 - Works from the URL Scanner as well https://t.co/3MpjvHT2YH https://t.co/jClbl1OoAM - **appfactory** (2026-04-17): https://x.com/appfactory/status/2045168017491800225 - @irvinebroque I just pasted it all in: https://t.co/YUqEPoADP1 https://t.co/WxQ7EMFIjz - **wongmjane** (2026-04-17): https://x.com/wongmjane/status/2045167774264447160 - Not me being featured as a jester who messes around Cloudflare’s feature flags 💀 https://t.co/N6XuhJRVxm https://t.co/6xRe1QbphJ - **Cloudflare** (2026-04-17): https://x.com/Cloudflare/status/2045167722343072019 - We are launching Flagship, a native feature flag service built on Cloudflare’s global network to eliminate the latency of third-party providers. https://t.co/gtu5xwSVCa - **appfactory** (2026-04-17): https://x.com/appfactory/status/2045166358502015474 - @irvinebroque already doubled my scores! - **michellechen** (2026-04-17): https://x.com/michellechen/status/2045159347831939357 - now im literally her https://t.co/gZ8ZGM1IwI https://t.co/STtnLOdrVO - **JessicaNutt96** (2026-04-17): https://x.com/JessicaNutt96/status/2045153272906891745 - Mythos is not a bad name for a model but it would be better if Anthropic switched to using famous Claudes. Monet, Debussy etc. The final model that achieves AGI would obviously be Van Damme - **fayazara** (2026-04-17): https://x.com/fayazara/status/2045150981604986918 - What can flagship do? 1. Roll out features gradually. 2. Hide features from @wongmjane https://t.co/QUv0WFRE0e - **roerohan** (2026-04-17): https://x.com/roerohan/status/2045140184329753019 - An AI agent can write your code in minutes. But someone still has to review, merge, deploy, and monitor it. What if the agent could do that too? Feature flags are the missing piece. They let an agent ship code behind a flag, test it on real traffic, ramp the rollout, and kill it instantly if things break. No human in the loop until you choose to be. Today we're shipping Flagship to make this possible - feature flags native to @Cloudflare's network, OpenFeature standard. Move fast, break nothing. https://t.co/UYWTNO182f - **fayazara** (2026-04-17): https://x.com/fayazara/status/2045133183575113771 - Today, we're announcing Flagship - Cloudflare's native feature flag service, built for the AI first world. Read more about it here https://t.co/wiYd606ksc - **Cloudflare** (2026-04-17): https://x.com/Cloudflare/status/2045132731831427556 - Soft directives don’t stop crawlers from ingesting deprecated content. Redirects for AI Training allows anybody on Cloudflare to redirect verified crawlers to canonical pages with one toggle and no origin changes. https://t.co/e3ByECc01v - **KankaniAbhishek** (2026-04-17): https://x.com/KankaniAbhishek/status/2045132371440062913 - Rollout complete!🏁 Introducing Flagship: @Cloudflare's native feature flag service built for the AI-first world. OpenFeature compatible, evaluated across region Earth🌐 Why we built it 👇 https://t.co/Fe4S6wK1VD https://t.co/d5PJysjHqO - **michellechen** (2026-04-17): https://x.com/michellechen/status/2045131092487815314 - that's my quant Unweight: a lossless compression system that can make model weights up to 15–22% smaller while preserving bit-exact outputs, without relying on any special hardware https://t.co/nc2LXNi4GA - **Cloudflare** (2026-04-17): https://x.com/Cloudflare/status/2045126394418503846 - Today, we are excited to introduce https://t.co/F5gmrAYGFP — a new tool to help site owners understand how they can make their sites optimized for agents. https://t.co/2xAeZlX5AI - **LakeAustinBlvd** (2026-04-17): https://x.com/LakeAustinBlvd/status/2045115948063989765 - someone in the Cloudflare Lisbon office trying to catch up on all the Agent Week blog posts, I guess https://t.co/fZ2LvFvGUv - **mrramibanna** (2026-04-17): https://x.com/mrramibanna/status/2045018855777665522 - We're building the Stripe Projects network at a rapid clip... come see: https://t.co/011iIiAmpd https://t.co/q3S0XPTPtk - **zeke** (2026-04-17): https://x.com/zeke/status/2044989164408754434 - These SVG models just get better and better. 🍿 Great post by @shridharathi https://t.co/QHraYkqApS https://t.co/Ak3Dg4XLkw - **JoeVirayNBA** (2026-04-17): https://x.com/JoeVirayNBA/status/2044968705172603075 - Marcus still the GOAT https://t.co/laFehkpxwe - **garrytan** (2026-04-16): https://x.com/garrytan/status/2044923648218452455 - Open source software will be many times more secure than closed source software in the new Mythos era https://t.co/a1edlJ7FEK - **dmtrubman** (2026-04-16): https://x.com/dmtrubman/status/2044899388204470711 - Such a bummer knowing left wing transit advocacy groups are going to spend the next few decades campaigning against autonomously driven buses (which could drastically increase service hours) because they prioritize the demands of transit unions, not the riding public. https://t.co/8pUrKpgMkI - **clairevo** (2026-04-16): https://x.com/clairevo/status/2044877250051580188 - this is one of my favorite AI stories lately. @intercom is just crushing it lately. and psst -- watch the How I AI channel monday if you want a peek behind the scenes... https://t.co/oISV2EQC7j - **stevekaliski** (2026-04-16): https://x.com/stevekaliski/status/2044873864887275795 - We're adding new providers to https://t.co/noiCP4o4JT as swiftly as possible including @huggingface, @Cloudflare, @OpenRouter, @firecrawl, @flydotio, @Amplitude_HQ, @mixpanel, and @inngest. Tell your agent to run `$ stripe projects catalog` to see them all. - **patocallaghan** (2026-04-16): https://x.com/patocallaghan/status/2044873364007694610 - Haven't written a line of code in 8 months. Don't even open an editor anymore. Yet shipped more than I ever have. @intercom's 2x initiative has been the wildest 9 months of my career. https://t.co/e5voGe6ZR9 - **zeeg** (2026-04-16): https://x.com/zeeg/status/2044871690275201450 - Sentry's MCP server now gives access to nearly all datasets within Sentry: Issues, Logs, Traces, Metrics (Beta), Profiles, and Session Replays Weve done a lot of optimization for debugging errors, but if you find yourself getting value out of those others, give it a spin! - **patrickc** (2026-04-16): https://x.com/patrickc/status/2044870698196513096 - Just launched some great new providers in Stripe Projects (https://t.co/QwQF6d2TIz): @huggingface @Cloudflare @OpenRouter @firecrawl @flydotio @Amplitude_HQ @mixpanel @inngest Many more coming later this month. - **jeff_weinstein** (2026-04-16): https://x.com/jeff_weinstein/status/2044865902420832394 - Just added 8 new developer-loved providers to @stripe projects: > stripe projects catalog > @huggingface, @Cloudflare, @OpenRouter, @firecrawl, @flydotio, @Amplitude_HQ, @mixpanel, @inngest Your agents can now prompt, provision, and pay all these providers, programmatically! :P https://t.co/ebhgEI24uZ - **kevinweil** (2026-04-16): https://x.com/kevinweil/status/2044862783947448611 - 💥 Super excited to launch GPT-Rosalind, our first frontier model built for scientific research across biology, drug discovery, and translational medicine. The model is trained in chemistry, protein engineering, genomics and more, including built-in knowledge of relevant databases and tools researchers use. We want to make these capabilities available to scientists and researchers while maintaining strong safeguards against biological misuse. GPT-Rosalind is launching through a trusted access deployment structure for qualified customers. We're also launching a Life Sciences plugin for Codex to everyone today, which works with our mainline models as well as GPT-Rosalind. - **sama** (2026-04-16): https://x.com/sama/status/2044858862042591378 - Lots of major improvements to Codex! Computer use is a real update for me; it feels even more useful than I expected. It can use all of the apps on your Mac, in parallel and without interfering with your direct work. - **stripe** (2026-04-16): https://x.com/stripe/status/2044857514798260336 - New providers now available in Stripe Projects: @huggingface @Cloudflare @OpenRouter @firecrawl @flydotio @Amplitude_HQ @mixpanel @inngest https://t.co/eDeHfCaAAG - **brian_scanlan** (2026-04-16): https://x.com/brian_scanlan/status/2044839059239973336 - https://t.co/eZeX3Le6gv Important update from Darragh - we hit our goal of doubling engineering throughput 3 months early. https://t.co/7JxTFu3jKg - **devahaz** (2026-04-16): https://x.com/devahaz/status/2044834824951849102 - SEIU should switch from wealth tax proposal to this - **thdxr** (2026-04-16): https://x.com/thdxr/status/2044833584947622110 - opus 4.7 is a beauty a fresh yet elegant take on something we've seen before a new standard, a definite marker of a new era (i haven't tried it yet) - **devahaz** (2026-04-16): https://x.com/devahaz/status/2044832908679274670 - In 2020 a measure to eliminate prop 13 protections for large commercial properties failed in a close vote getting 48%. I think that measure would pass today and should be extended to eliminate prop 13 for all properties other than owner-occupied primary residences. https://t.co/EC1I1XxMZP - **vimtor** (2026-04-16): https://x.com/vimtor/status/2044829991788036202 - now you can deploy @astrobuild on cloudflare this includes custom domains, resource linking, and local support for bindings during sst dev https://t.co/AuiD8uEvVh - **brian_lovin** (2026-04-16): https://x.com/brian_lovin/status/2044829132442435733 - wtf is in the water at cloudflare? - **thsottiaux** (2026-04-16): https://x.com/thsottiaux/status/2044826325173879269 - Codex just got a lot more powerful. Computer use, in-app browser, image generation and editing, 90+ new plugins to connect to everything, multi-terminal, SSH into devboxes, thread automations, rich document editing. Learns from experience and proactively suggestions work. And a ton more. - **ritakozlov** (2026-04-16): https://x.com/ritakozlov/status/2044824092814053869 - linkedin recruiter needs a filter out people with several pipes in their bio option - **darraghcurran** (2026-04-16): https://x.com/darraghcurran/status/2044821428453646632 - 9 months ago we publicly committed to 2x the productivity of our R&D org at @intercom. It was scary. It wasn't always clear we'd pull it off. We hit it with 3 months to spare. In fact, looking back 16 months - we've 3x'd. Here's what actually happened (with receipts): 🧵 https://t.co/vYGlGnVOGB - **chrismunns** (2026-04-16): https://x.com/chrismunns/status/2044791648316449010 - More on our partnership with @Cloudflare. Really exciting architecture patterns are forming and I'm really excited that @PlanetScale databases get to be a part of it, from $5 DBs up through well.. we haven't found a top yet... https://t.co/HVG23VLeP8 - **minglu** (2026-04-16): https://x.com/minglu/status/2044790778149605593 - i’m pretty sure irtefa set a record here for most number of partners gotten through legal and procurement in the shortest amount of time. https://t.co/Q393wmxWgp - **_irtefa** (2026-04-16): https://x.com/_irtefa/status/2044787604957077716 - 💌💌💌 https://t.co/wY9ZqUJcga - **minglu** (2026-04-16): https://x.com/minglu/status/2044784785298788674 - excited to debut on the cloudflare blog with what we’ve been cooking over on ai gateway: https://t.co/WFCwh2QvKV run the best models, pay with one bill, and see it all in one place. lots more to do! - **WalshyDev** (2026-04-16): https://x.com/WalshyDev/status/2044783662319702072 - We finally did it! Email Sending is available for customers!! This has been one of the biggest feature requests for us for a very long time, super super happy it is now available Huge congrats to the team for shipping this! https://t.co/c4M0Zk9F6h - **dok2001** (2026-04-16): https://x.com/dok2001/status/2044782212457824403 - This easy. https://t.co/Hm42QrYkRv - **LambrosPetrou** (2026-04-16): https://x.com/LambrosPetrou/status/2044779449766252869 - You really can just build anything these days on Cloudflare Workers and Durable Objects. https://t.co/7DhnmQsZuT - **eastdakota** (2026-04-16): https://x.com/eastdakota/status/2044778863922655390 - New tools for a new world. https://t.co/c1aOpx4orA - **_ashleypeacock** (2026-04-16): https://x.com/_ashleypeacock/status/2044778148194062596 - You can import a repo straight from GitHub into Artifacts Then just clone it as you normally would https://t.co/jE3QJYhUBM https://t.co/0V3EyyL5qZ - **fayazara** (2026-04-16): https://x.com/fayazara/status/2044775207551025189 - AI Gateway is now a unified inference layer. You can call 50+ models across 12+ providers through the same https://t.co/xxsHQdacGx() binding as Workers including new image, video, and speech models. Get logs, spend management and billing all in one place https://t.co/68b3PQYGJ7 - **aninibread** (2026-04-16): https://x.com/aninibread/status/2044773045772959944 - We shipped big changes for @CloudflareDev AI Search! Every agent needs search and it shouldn't take a week to set up. AI Search is the search primitive for your agents. Create an instance, upload your data, and search it. Comes with hybrid retrieval with vector and BM25. https://t.co/mKgnfAPjRk - **ritakozlov** (2026-04-16): https://x.com/ritakozlov/status/2044772015005966778 - it’s been incredible working on this with the @replicate team! @minglu @corywilkerson @superhighfives @mattrothenberg and so many more 👏 incredible team! - **michellechen** (2026-04-16): https://x.com/michellechen/status/2044771923398181203 - cog makes containerizing machine learning models dead simple: all you need is a 𝚌𝚘𝚐.𝚢𝚊𝚖𝚕 and a 𝚙𝚛𝚎𝚍𝚒𝚌𝚝.𝚙𝚢 building the path to bring your own model on workers ai https://t.co/4t8A3nxbvI https://t.co/slMXOjicQo - **ritakozlov** (2026-04-16): https://x.com/ritakozlov/status/2044771567800865210 - cloudflare is the ai cloud you need ⛅️ you can now access any model including gpt-5.4, opus 4.6, kimi, seedream, nano banana, … anything you need with a single api call, no vendor lock-in and we’re constantly adding more! tell us what models you’d like to see next! https://t.co/PLqmkH1ZSb - **_irtefa** (2026-04-16): https://x.com/_irtefa/status/2044770281781105028 - Today the average company calls 3.5 models across multiple providers. That means 3.5 billing dashboards, 3.5 sets of API docs, and no single place to see what you're actually spending. When one provider goes down, you're writing custom failover logic at 2am. We just shipped something at Cloudflare that changes this. Today we shipped AI Gateway as a unified inference layer at Cloudflare: one API, one line of code to switch between any model from @alibaba_cloud , @AssemblyAI , @BytedanceTalk , @Google , @inworld_ai , @MiniMax_AI , @Kimi_Moonshot , @OpenAI , @PixVerse_ , @recraftai , @runwayml , and @ViduAI_official . Seventy plus models (and growing!), twelve plus providers, one set of credits, automatic failover when a provider goes down. None of this works without partners who chose to make their models accessible through a shared, open platform instead of walled gardens. Grateful to every one of them for building this with us. - **dok2001** (2026-04-16): https://x.com/dok2001/status/2044769850111734186 - Agents know Git. So we gave them Git. Artifacts: versioned storage built for agents, speaking a protocol they already understand. https://t.co/htY9jIl0aa - **celso** (2026-04-16): https://x.com/celso/status/2044768831055204845 - AI Search now comes with its own storage and vector index, fully managed and adds hybrid search with both semantic and keyword matching. https://t.co/0I3g6xGcSY - **dok2001** (2026-04-16): https://x.com/dok2001/status/2044768597109510421 - Email is older than the web. It’s also the best interface for agents. Everyone already has an address. No app to install. No SDK to integrate. Cloudflare Email Service is now in public beta. https://t.co/NSHGfuwM0o - **ritakozlov** (2026-04-16): https://x.com/ritakozlov/status/2044768407346688044 - we run unquantized kimi 2.5 at massive scale. we made a series of optimizations to make it 3x (!!) faster here’s how 👇 https://t.co/2WnOU6PFos - **michellechen** (2026-04-16): https://x.com/michellechen/status/2044768144862990545 - 𝚎𝚗𝚟.𝙰𝙸.𝚛𝚞𝚗("𝚘𝚙𝚎𝚗𝚊𝚒/𝚐𝚙𝚝-𝟻.𝟺") one binding to hit all models, including proxied models like gpt-5.4 or nano banana, and hosted models on workers ai. one more step towards unification for workers ai, ai gateway, replicate https://t.co/4t8A3nxbvI - **celso** (2026-04-16): https://x.com/celso/status/2044767253141303720 - Incredibly proud of the team behind this. Email Sending enters public Beta today. https://t.co/3ST8UFcT14 - **elithrar** (2026-04-16): https://x.com/elithrar/status/2044767190834991490 - We just shipped “Git for agents”. Turns out agents are really good at working with Git, but existing source control platforms weren’t built for the volume of commits we’re seeing now. Create tens of millions of repos. Use them from any Git client. https://t.co/PNxOFErLGl - **whoiskatrin** (2026-04-16): https://x.com/whoiskatrin/status/2044767066440225228 - cloudflare just gave agents git this is one of those changes that will just quietly improve everything agents with proper version control @dillon_mulroy @elithrar @mattzcarey @thomas_ankcorn have done something incredible here https://t.co/4dFPie896A - **thomasgauvin** (2026-04-16): https://x.com/thomasgauvin/status/2044766954032951792 - The wait is over. Cloudflare Email Service is now in public beta 📧 Send and receive emails directly from Workers or REST API with global delivery on Cloudflare's network And just in time for you to build email agents with the Agents SDK! https://t.co/4pyyjWjsLm - **michellechen** (2026-04-16): https://x.com/michellechen/status/2044766865348575363 - serving models is hard. serving extra large language models with good quality, throughput, reliability, pricing, and gpu utilization is really hard. a look behind the scenes of how we do it on workers ai this agents week. there’s always more to do, but we’re just getting started https://t.co/Cs20pf4fBG - **ZelenskyyUa** (2026-04-16): https://x.com/ZelenskyyUa/status/2044766437806665792 - Today in Ukraine is another very hard day, really hard night – the day after a massive Russian attack. In just one night, there were nearly 300 attack drones, 19 ballistic missiles, as well as cruise missiles. Air-raid sirens sounded across many of our cities. And this came after waves of “shaheds” had filled our skies just yesterday. Dozens of people have been injured. And sadly, so sadly, there are also lives lost in Odesa, Kyiv, in Dnipro. Just ordinary people, children, civilians, killed by Russian madness. If it’s possible, please, I ask you now to honor their memory, and all those whose lives have been taken by this terrible war, with a minute of silence. Thank you. From my address to the participants of the Four Freedoms Awards ceremony (1/3). - **zebassembly** (2026-04-16): https://x.com/zebassembly/status/2044766139192942940 - We're launching Git storage for Agents built on Durable Objects and Zig! https://t.co/avIe8kONzF https://t.co/oXxDOEZkHb - **dillon_mulroy** (2026-04-16): https://x.com/dillon_mulroy/status/2044765430649168272 - announcing artifacts - a versioned file system that speaks git certainly the coolest thing i've ever built and i got to build it with some truly great people fearless leader: @elithrar zig/git nerd: @mattzcarey storage goat: @thomas_ankcorn https://t.co/BSjfaXTnvJ https://t.co/Pk3wNjtbYk - **theo** (2026-04-16): https://x.com/theo/status/2044697225666929151 - The Claude Code Desktop app is an affront on software. As developers, we should be offended that they chose to ship something this awful. Rushed out my video because I feel like I'm going insane. https://t.co/kWdszY1oDJ - **warriorsworld** (2026-04-16): https://x.com/warriorsworld/status/2044662586885767662 - https://t.co/Y6En9mFFVH - **anthonyVslater** (2026-04-16): https://x.com/anthonyVslater/status/2044661736696168911 - Draymond Green: “The same people who say I lost a step say I never had a step.” https://t.co/3aM4AX8OqY - **JoeVirayNBA** (2026-04-16): https://x.com/JoeVirayNBA/status/2044654509436346786 - I will not let this ATO get lost amidst the Steph/Draymond/Horford craziness. With the Clippers content on "hiding" Garland on Gui Santos, Steve Kerr draws this play up for Gui. Split action variant where Gui is supposed to set the rip screen for Podz. But the Gui lift to supposedly set the screen is a disguised way to get Garland in the paint on a seal. Garland had five fouls at this point and is giving up 6 inches in height. Incredible belief by Kerr in Gui which he repays (in more ways this game than just this sequence). - **JoeVirayNBA** (2026-04-16): https://x.com/JoeVirayNBA/status/2044638526235643995 - DRAYMOND GREEN IS THIS GENERATION'S GREATEST DEFENDER IDC https://t.co/PbQdWRU0H7 - **samlambert** (2026-04-16): https://x.com/samlambert/status/2044592666328113550 - incredible company https://t.co/uR3uoxPdHB - **ThompsonScribe** (2026-04-16): https://x.com/ThompsonScribe/status/2044585974332837993 - Moses Moody was built differently. Even for the hardest of times https://t.co/lSc2E4msd2 https://t.co/Fn7vp7buNu - **pmarca** (2026-04-16): https://x.com/pmarca/status/2044585024360444327 - Nope. Nope nope nope nope nope nope nope https://t.co/Z40mIJ09fa - **bouhmad** (2026-04-15): https://x.com/bouhmad/status/2044503123813949923 - PM on the @Cloudflare IAM AuthZ team here👋🏾 We've shipped a lot of IAM improvements at @Cloudflare over the past year, and have a lot more planned. What's still frustrating? What would make your life easier? DMs are open https://t.co/8YdWC0fvq5 - **FredKSchott** (2026-04-15): https://x.com/FredKSchott/status/2044479150489276896 - OpenAI is using the term "Sandbox Agent" now. Wild. When I created Flue I'd envisioned this idea of sandbox agents as just a CI-only niche, but clearly this is now where the entire industry is headed. Any agent can be generalized to: - A file system: to define context, skills, etc. - An agent: to capture instructions, role, etc. - A harness: to run the agent - A sandbox: to provide the runtime. Note that this can run in a full container or a faster/cheaper isolate (microVM, in-memory FS, Cloudflare DO, just-bash, etc) depending on the problem you get it. In the future every agent will be represented this way. It no longer makes sense to learn a bespoke agent framework in python or JS when every dev already knows how to set up Claude Code, Codex, OpenCode. Should Flue become a new agent framework that lets you build any agent, the same way to setup Claude Code? https://t.co/SI2lWF51Me - **CloudflareDev** (2026-04-15): https://x.com/CloudflareDev/status/2044467412607901877 - Long-running agents are the future - we're excited to partner with OpenAI as a sandboxing partner for their new Agents SDK launch! Get started: https://t.co/OoiBR5S05l https://t.co/g2UAnTSSPK - **aninibread** (2026-04-15): https://x.com/aninibread/status/2044450595080110318 - Also, you can now add the ability to talk to your agents. @Cloudflare /voice adds real-time voice to the Agents SDK. ~30 lines of server code to get started, with built-in speech to text and text to speech on Workers AI. https://t.co/X1l0CLLoLX - **BraydenWilmoth** (2026-04-15): https://x.com/BraydenWilmoth/status/2044422996765352226 - Cloudflare dashboard can now complete tasks for you. - "Create a Worker and bind a new R2 bucket to it" - "Change my DNS records to 1.1.1.1" - "How many errors have happened this week" Not only do we tell you, but we show you with generative UI. PROTIP: Use full-screen mode. https://t.co/Q1o1vyoOwk - **EastlondonDev** (2026-04-15): https://x.com/EastlondonDev/status/2044419711690055957 - Cloudflare are shipping everything I wantedwhen I was working deep on the platform ~6 months ago. Everything that was missing from mesh networking to dynamic workers to the agents framework data structures to easily using R2 with agents is now here - **whoiskatrin** (2026-04-15): https://x.com/whoiskatrin/status/2044415568627847671 - today project Think is officially out! we bet on agents that run non-stop, survive failures, cost nothing when idle, and enforce security through architecture agents that any developer can build and deploy agents that have sub-agents via Facets, Session API and full execution ladder wired in! https://t.co/CwXvWHX2mE - **aninibread** (2026-04-15): https://x.com/aninibread/status/2044409784133103724 - Project Think is here. It's the next generation of the @CloudflareDev Agents SDK with both lightweight primitives to a full suite of tools that you can use to build long-running agents. Durable execution, sub-agents, persistent sessions, sandboxed code execution, a built-in workspace filesystem, and agents that can even write their own tools at runtime. It's available in preview now. - **thearmadalabs** (2026-04-15): https://x.com/thearmadalabs/status/2044403949403263160 - Builders have been asking @Cloudflare for a registrar API for a while. Good news, for agents week we are launching in beta! https://t.co/ppsKSYTjf8 Search domains, check availability, a buy! Still at-cost, no markup. Can’t wait to see what y’all build. - **DanielleMorrill** (2026-04-15): https://x.com/DanielleMorrill/status/2044243902551732650 - anyone else noticing how fast Cloudflare is shipping? - **ctatedev** (2026-04-15): https://x.com/ctatedev/status/2044224002109907392 - Introducing wterm (“dub-term”) A terminal emulator for the web → DOM rendering — not canvas → Select text, copy/paste, ⌘+F, a11y → Dirty-row tracking, 24-bit color, themes → WebSocket transport with reconnection → Zig core compiled to ~12 KB WASM → just-bash, local, SSH https://t.co/btV6ph5fyB - **parkerconrad** (2026-04-15): https://x.com/parkerconrad/status/2044221628343824717 - Rippling AI was the most successful launch we've ever done. On the heels of this launch, Rippling's revenue is now growing 78% YoY (at ARR over $1 Billion). And this growth rate has now increased, every quarter, for three straight quarters. https://t.co/nLPtMYtQOj - **akothari** (2026-04-15): https://x.com/akothari/status/2044214909270200618 - You can use AI to cut costs. You can also use AI to raise your ambition. Too many people focus on the first. More people should be thinking about the second. - **davidhoang** (2026-04-14): https://x.com/davidhoang/status/2044173195465634074 - In addition to AI, I now lead Jira Design. There is a huge opportunity ahead, but requires much work. Many designers talk about wanting to work on tough problems and impact, but few are willing to take it head-on. I'm grateful for the opportunity. - **elithrar** (2026-04-14): https://x.com/elithrar/status/2044172401454223823 - I was at a customer workshop a couple of weeks ago w/ a team that has been building on us for a decade. (paraphrasing) “we know what the other clouds will look like in the next 5-10 years. They’ll look mostly the same. We know that’s not going to be the case with Cloudflare.” https://t.co/vENYFyJJHT - **dillon_mulroy** (2026-04-14): https://x.com/dillon_mulroy/status/2044158080288506207 - i rest my case https://t.co/7w74hogNT5 https://t.co/XQFxrLd3ek - **dok2001** (2026-04-14): https://x.com/dok2001/status/2044130733904581000 - @Cloudflare x @OpenAI . GPT-5.4 and Codex, running on Dynamic Workers, and Sandboxes. Agents need a home. This is it. https://t.co/pA8Ok3igzu - **burcs** (2026-04-14): https://x.com/burcs/status/2044122398916542604 - some non-agent launches shipped today too... we've simplified the api token creation and member invite flows! try it out and let me know what you think 🧡 https://t.co/tY3tpt7go6 - **destraynor** (2026-04-14): https://x.com/destraynor/status/2044101003876655504 - Thrilled to see this one ship to everyone. It's one thing to hand roll an AI feature once for one customer. Rolling this out reliably to 8000 customers is a different type of challenge. Kudos to Pratik & Team https://t.co/TRP87fPEDn https://t.co/g3vB71F50d - **pratik60** (2026-04-14): https://x.com/pratik60/status/2044098287217586520 - Fin Procedures is now available to every @intercom customer - all 8,000+, no waitlist. Complex queries are a small percentage of your support queue but a huge percentage of your team's time. Most AI agents handle the easy stuff. For the hard stuff, you're told you need an FDE and weeks of setup. - **ritakozlov** (2026-04-14): https://x.com/ritakozlov/status/2044089959536087078 - lots of stuff in this blog post alone, so a little summary 🧵 tldr: you can have your cake and eat it too... errr... vibe code things / let agents run wild but do it securely read on - **mattzcarey** (2026-04-14): https://x.com/mattzcarey/status/2044072842606555566 - easiest way to do code mode just dropped. https://t.co/PO4pN9Btpx https://t.co/OG193n47Qe - **mattzcarey** (2026-04-14): https://x.com/mattzcarey/status/2044055851036483801 - Sharing some thoughts on MCP for enterprise and releasing Code Mode for MCP Portals. https://t.co/3ufwUuKh8e - **chrismunns** (2026-04-14): https://x.com/chrismunns/status/2044049149109125229 - It really looks like @Cloudflare is building the Cloud 2.0, direct for cloud-first/digital native companies/AI-first. This isn't a legacy cloud, its a today+future building one. its very different from where "legacy" cloud is today. - **elithrar** (2026-04-14): https://x.com/elithrar/status/2044047674861654487 - you have hundreds or thousands of agents, that need access to data & services across your cloud infra & SaaS providers. some of those agents will need to talk to other agents. Mesh is the foundation of what enables orgs to ship secure agents that do not leak your data. https://t.co/Lw9KHwx4rg - **aninibread** (2026-04-14): https://x.com/aninibread/status/2044044739616399567 - Day 2 - here we come! Get excited about Cloudflare Mesh. This is how you run OpenClaw better. Your agent is on a Mac mini at home and you want to reach it from your phone. Exposing it to the public Internet is risky. Mesh gives it a private network so you can securely connect from any device, routed through Cloudflare. - **thomasgauvin** (2026-04-14): https://x.com/thomasgauvin/status/2044039230595211761 - Cloudflare Mesh is here. Ready to connect your devices, servers, and agents to a single private network 🔐 And with Workers VPC, your Workers, Agents, and Durable Objects running on Cloudflare can now reach your private MCPs, APIs, and databases directly Oh, and it's 50 nodes + 50 users free on every account - **LakeAustinBlvd** (2026-04-14): https://x.com/LakeAustinBlvd/status/2043948071134871589 - this is a very very good write up not enough time being spent right now thinking clearly like this about AI outside of R&D I think this post nails it. start with the job-to-be-done, rethink the factory, and empower super operators https://t.co/4SgEXzeBg3 - **thenanyu** (2026-04-14): https://x.com/thenanyu/status/2043885103470805376 - @levie We’re reinventing product managers from first principles - **bryan_johnson** (2026-04-14): https://x.com/bryan_johnson/status/2043863307111575677 - Friends, stop drinking alcohol. Not cut back. Eliminate. > alcohol increases cortisol > disrupts REM sleep > accelerates epigenetic aging > shrinks hippocampal volume > elevates resting heart rate > raises inflammatory markers > impairs glucose metabolism for 16 hrs One drink does that. - **badlogicgames** (2026-04-13): https://x.com/badlogicgames/status/2043817891053875648 - People of pi. New release now includes this ping to https://t.co/TgG5bkYs3t for the reasons outlined below. You can opt-out via any of these options: settings.json -> enableInstallTelemetry: false /settings PI_TELEMETRY set to 0/false PI_OFFLINE also includes disablement https://t.co/TgG5bkYs3t only gets the ping in interactive mode. RPC, SDK, print, or JSON mode are not affected. The ping only updates a counter on the server. The server does not store any data about you, your machine, or anything else you can think off. Not even your ip. Thanks for your help. - **reed** (2026-04-13): https://x.com/reed/status/2043765835190415415 - @Eric_Erins @tnertz that’s what’s cool about waymo, even if you aren’t using them they make you safer - **EricNewcomer** (2026-04-13): https://x.com/EricNewcomer/status/2043757435194667319 - sign me up for whatever the opposite of this is... oh right... safer rides from an amazing feat of human engineering https://t.co/AWt6YaQySd - **kane** (2026-04-13): https://x.com/kane/status/2043737683575615493 - https://t.co/JgVIUiy3zC https://t.co/emnXIWcMzo - **avenceslau** (2026-04-13): https://x.com/avenceslau/status/2043731920853324213 - For all the workflows users out there come see the beautiful local workflows dash by our own @opombosilva You can also do restarts, pause, etc, all the great apis from prod directly on your local dev setup Let us know of any problems you still have in local dev we will fix them - **reed** (2026-04-13): https://x.com/reed/status/2043731353892728991 - @MorePerfectUS being anti waymo is like being anti cancer cure. should we stop cancer research to save chemotherapist jobs? - **YIMBYLAND** (2026-04-13): https://x.com/YIMBYLAND/status/2043720828391473178 - If Waymo gets its way, tens of thousands of people will be spared from automotive-related deaths. When Waymo gets a firm hold on a city, pedestrians, pets, and drivers are all safer. This isn’t inevitable — The pro-auto-fatality lobby is doing everything possible to stop it. https://t.co/lXuEtPyRiX - **elithrar** (2026-04-13): https://x.com/elithrar/status/2043709964443619551 - https://t.co/pOE1UiIMRm https://t.co/dHx5wD9VPc - **nurodev** (2026-04-13): https://x.com/nurodev/status/2043707764246024584 - So proud of my first major feature shipping! 🚢 You can finally view and edit your local Wrangler state in either a clean UI or your agent to use the built-in API https://t.co/zrS09XFyXC - **adamkovac** (2026-04-13): https://x.com/adamkovac/status/2043706387776098810 - More Perfect Union taking a bold pro-road-fatalities stance https://t.co/da3nP9aSFC - **micsolana** (2026-04-13): https://x.com/micsolana/status/2043700673435603087 - why the fuck are we sharing images of sam's home in the press? at this point in time, in coverage of what looks like may have been a *second* attack, how could you possibly justify this? - **MikeNomitch** (2026-04-13): https://x.com/MikeNomitch/status/2043696888105570454 - Cloudflare Sandboxes now do secure credential injection. We use Workers as a trusted proxy, which make it really easy to customize egress policies (globally or per-instance) and update them dynamically. Let's take a look... Inject a token without the agent ever seeing it: https://t.co/zoT1AmGN2r - **scottew** (2026-04-13): https://x.com/scottew/status/2043690237763957065 - If Henry Ford gets his way, 2 million horse and buggy drivers will be out of work. When Ford gets a firm hold on a city, wages go down. Some buggy drivers now have to work 12 hours a day, 7 days a week just to get by. https://t.co/3Bof6E2OJT - **ritakozlov** (2026-04-13): https://x.com/ritakozlov/status/2043688018578489388 - if @KentonVarda says you should read it, you probably should this also might be one of the best compliments i have ever received 🧡 https://t.co/4WWHoloWzc - **CFchangelog** (2026-04-13): https://x.com/CFchangelog/status/2043684671104450905 - Containers and Sandboxes are now generally available. Run heavy workloads, Docker images, and SSH for debugging. https://t.co/qBKUXXnlPt - **KentonVarda** (2026-04-13): https://x.com/KentonVarda/status/2043684025454170438 - A fun new lego piece for your Cloud Agents set: Durable Object Facets. https://t.co/ypdid1Q2rt When sandboxing agent code with Dynamic Workers, you can now give the sandbox access to a SQLite database. The sandboxed code accesses this by literally just writing a Durable Object implementation, the same way any DO would be written. The full SQLite API is available (with fast synchronous reads/writes). Importantly, you DON'T have to create a regular DO namespace to attach to the sandboxed code. Instead, you provide the specific database for any particular instance to use. Where do you get this database? Your own Durable Object, acting as a supervisor, is now able to create one or more attached databases, which in can then pass into Dynamic Workers. We call these "facets", since all the databases are part of the same Durable Object. A bit wonky but it works out really nice. Previously, you would have had to give your Dynamic Worker an RPC interface that talks to some remote storage, which is a lot slower and forces async I/O which can lead to concurrency bugs. More importantly, your RPC interface would be unfamiliar to the agent, whereas now it can just write against the standard Durable Object storage APIs that all models already have in their training set. - **sarthakgh** (2026-04-13): https://x.com/sarthakgh/status/2043680370286096403 - This is just really bad for our future https://t.co/xPsKzVEKdH - **vimtor** (2026-04-13): https://x.com/vimtor/status/2043625288819429853 - this week i’ll be improving @cloudflare support in sst please open an issue with any problems or features you might want - **Noahpinion** (2026-04-12): https://x.com/Noahpinion/status/2043466594886811748 - Yes, it's regulatory. https://t.co/djHIaXFnZk https://t.co/2tA78P6sGt - **cremieuxrecueil** (2026-04-12): https://x.com/cremieuxrecueil/status/2043425845247639790 - I can hate this! The property tax is one of the best taxes on the books and it's pro-family to boot. Property taxes are capitalized into the cost of homes, making it easier to purchase them for young people looking to start families. Plus, they're efficient. Triple them! https://t.co/miq5ygmW2F - **dok2001** (2026-04-12): https://x.com/dok2001/status/2043380655770239062 - The Internet wasn't built for AI. The cloud wasn't built for agents. Every app before this was one-to-many. Agents flip it. One user, one agent, one task. This week we're shipping what that actually requires. https://t.co/LIKP7gJEDf - **ritakozlov** (2026-04-12): https://x.com/ritakozlov/status/2043375602128834711 - the internet wasn't built for the age of AI. and neither was the cloud tomorrow we're kicking off our first ever agents week, dedicated to building for what comes next 🚀 this might be the best innovation week cloudflare has ever had. here's why → https://t.co/I4uw7XX1RB - **lessin** (2026-04-12): https://x.com/lessin/status/2043370537280532771 - A bunch of people have written me back saying this was the best newsletter I have ever sent (flattering) ... so here it is for those who don't subscribe: AI Is Not a Labor Crisis. It Is a Meaning Crisis. - **zebassembly** (2026-04-11): https://x.com/zebassembly/status/2043083853024071890 - We're streamlining the Cloudflare leak experience and posting screenshots before @wongmjane can https://t.co/IKtXMbBgWh - **KentonVarda** (2026-04-11): https://x.com/KentonVarda/status/2043064767624155635 - Note that coalescing of writes is a separate thing from output gates. Writes that occur during a single visit to the isolate are automatically coalesced into a single atomic transaction. I just made up this term "visit" since we don't actually have a good term for this, but what it means is: we deliver some I/O event to the isolate, then we run the microtask loop until it is empty. This is all one "visit". Since we have our own runtime, yes, we are able to instrument the event loop to work how we want. We actually have two event loops: the microtask loop (which is specific to an isolate), and the outer KJ event loop (C++ code, using epoll, one per thread). Output gates, meanwhile, are implemented by instrumenting each type of I/O, such as fetch(). - **KentonVarda** (2026-04-11): https://x.com/KentonVarda/status/2043059511913562471 - Hello! I'm the guy to blame for all this input/output gate stuff (and durable objects generally). The intent here is you don't actually have to think about it basically ever. It's just an invisible force that works to keep your program behavior deterministic and correct. Input gates don't even matter that much anymore as storage operations are now synchronous ever since we switched to sqlite as the storage backend. Find more about synchronous storage APIs here: https://t.co/g12TrBI4RG If you use the sync APIs then *obviously* no other event can arrive in the middle of a storage call and input gates just don't matter anymore. As for output gates, you really don't have to think about them either. It's basically a speculative execution optimization, the same thing your processor does but you never thought about (er, until Spectre... never mind that). Semantically, when you write to storage, it either succeeds, or the whole durable object crashes and restarts. Storage failures are very rare, so much so that there's no reason why you should want to write code in your app to handle them. Just crashing and restarting is fine. What output gates do is allow your app to continue executing *speculatively* while the storage write is in progress. But if your app tries to communicate with the outside world, then we stop and wait to verify the storage write actually succeeded, before we allow the message to go out. This way, if the storage write fails, the outside world never knows that your app executed any code beyond that -- it appears the same as if the app just crashed on the write operation itself. What this means for you is you don't have to worry that doing several writes in series might be slow due to round trips. You don't wait for round trips. They all happen concurrently and you only wait when you send a reply to the client. - **michellechen** (2026-04-11): https://x.com/michellechen/status/2043040988868887032 - meow ^-^ https://t.co/xdRBaZWGVO - **boristane** (2026-04-11): https://x.com/boristane/status/2043034226912825791 - the closest to a programmable cloud is @cloudflaredev could be a tag line - **nandafyi** (2026-04-11): https://x.com/nandafyi/status/2043013002056929546 - location hint https://t.co/A4iCwTZoFZ - **threepointone** (2026-04-11): https://x.com/threepointone/status/2042966311374868605 - probably the first cloudflare/shell implementation I'm seeing in prod outside cloudflare a full vfs on top of do/r2, no bash shenanigans, just code inside a dynamic worker chufffffed https://t.co/1Dr9JAPpNm - **DianeSwonk** (2026-04-11): https://x.com/DianeSwonk/status/2042786037877870978 - Asked if I have ever seen anything like this labor market. NO. NOTHING. IT IS UNIQUE. https://t.co/Qv2CaWC7I2 - **levie** (2026-04-11): https://x.com/levie/status/2042759653281456218 - The more I meet enterprise CIOs and AI leaders outside of tech, the more it’s obvious that if you’re building software that doesn’t have a great headless mode, you’re going to be at risk in the coming years. Asked a group of 20 IT leaders across banking, media, finance, and healthcare if they will have any vendors left in 3-5 years that don’t have a good API option for their service and it was a unanimous “no”. This is clearly going to change the nature of software going forward. You have to be completely comfortable serving up your value proposition as much through agent on or off your platform, as you are your own interface. I suspect most platforms will make it to the other side because of how forceful the trend will be, but of course some won’t if their heads in the sand. But on the other end, the upside is that in a world of 100X+ more agents doing work with with software than people ever did, there are far more use-cases to drive and be a part of. In many ways it’s a renaissance if you’re tied to critical data or workflows because of what customers can now use you for. It will certainly force an evolution of business models over time - whether you embed all of this agentic usage in a seat license or make it all consumption based - but dollars will always flow to where value is created. Going to be fun! - **ruifigueira** (2026-04-10): https://x.com/ruifigueira/status/2042752465133535386 - Browser Rendering with CDP is so powerful that we even considered not releasing it to the public https://t.co/vJzgVPBfhx - **JoshuaKushner** (2026-04-10): https://x.com/JoshuaKushner/status/2042749545784778837 - sam is a wonderful human. many have worked very hard to make you think otherwise https://t.co/eZ5Wu1XtxT - **sama** (2026-04-10): https://x.com/sama/status/2042738954550603884 - I wrote this early this morning and I wasn't sure if I would actually publish it, but here it is: https://t.co/7Dw9UFpeep - **kathyyliao** (2026-04-10): https://x.com/kathyyliao/status/2042683531852419272 - your coding agents can now use Browser Rendering as a remote browser CDP is how browser automation works under the hood, so now your agents get max control over the browser https://t.co/ZWQ9V2AtQo - **CloudflareDev** (2026-04-10): https://x.com/CloudflareDev/status/2042678424444063865 - Browser Rendering now exposes the Chrome DevTools Protocol, which means you can give your MCP clients access to a remote browser. Update your MCP config and your agent can navigate pages, take screenshots, run performance audits, and debug JavaScript. https://t.co/gj4zn7Sp4K - **grinich** (2026-04-10): https://x.com/grinich/status/2042661817550541253 - The UI era is ending. 🪦 For 70 years we designed computer interfaces. Mainframe, CLI, GUI, Touch. But with AI, the interface is disappearing. What will come next? My talk from @mastra's conf this week: https://t.co/VBDoVAzsbV - **threepointone** (2026-04-10): https://x.com/threepointone/status/2042653919038304663 - @abuiles @irvinebroque here you go https://t.co/B3hbaDPJyO - **abuiles** (2026-04-10): https://x.com/abuiles/status/2042653433598022129 - @irvinebroque Please tell me you guys are announcing an fs (or just bash) on top of DOs. I'm building a coding agent with PI and ended up hacking my own version + dynamic workers to let the LLMs get creative cc @threepointone https://t.co/bQfY3a678z - **MattieTK** (2026-04-10): https://x.com/MattieTK/status/2042632286768623868 - There's a great @EmDashCMS roundup from @emroth08 in The Verge today. If you've been looking for a bit of wider reaction from the blogging community, there's a bunch of super feedback and posts linked here. https://t.co/IlZPf95iFn - **levie** (2026-04-10): https://x.com/levie/status/2042469771275329594 - I think everyone is substantially underestimating the total demand for software and automation in areas that don’t feel like “software”. Not talking about software that’s another app on your phone. Software that just automates things for companies all day long. Most companies have not been able to bring automation to most areas of work because it’s been too complex or costly to do so. Outside of tech or maybe large banks, companies don’t have an unlimited supply of engineers. They have to ration resources very selectively, which means most things don’t get funded. Further, there are many projects that never got done simply because the technology wasn’t there to solve the problem. Basically anything even remotely touching unstructured data was impossible to automate before AI, or connecting data flows between systems that deal with significant variability, for instance. This is where all the new software and automation will be applied. It will be in CPG and retail connecting marketing stacks. Pharma research is about to explode because we can automate far more tests and simulations. Bankers and investors are going to run 10X the amount of analyses on every scenario. Healthcare providers will bring automation to the every step of the process. Now that agents bring down the cost of doing this work, and because many parts of it are now possible, companies will light these projects up. This is why there demand will continue for anyone technical enough to execute this work, and why the jobs arguments will be wrong. - **theo** (2026-04-10): https://x.com/theo/status/2042456180694663540 - “OpenAI is not shipping much” lmaooooo https://t.co/vPecBQUuNz - **joannejang** (2026-04-09): https://x.com/joannejang/status/2042379417679512042 - i knew openai labs was in the right hands when keyan shared this with me (congrats & thank you to @keyanzhang for taking on such a special, talent-dense team!! 💕) https://t.co/zaWinbcP8K - **brian_scanlan** (2026-04-09): https://x.com/brian_scanlan/status/2042358502304223296 - Doing live demos with agents is non-trivial. Here is a spectacular screenshot from one of the outtakes of a video I recorded about the cli. https://t.co/hTTF8qrTrA https://t.co/i15YqhFvEH - **michellechen** (2026-04-09): https://x.com/michellechen/status/2042351208883454122 - we made kimi 3x faster on @Cloudflare workers ai. more to come. enjoy! https://t.co/gu1rMlScb9 - **darraghcurran** (2026-04-09): https://x.com/darraghcurran/status/2042324505746350418 - Fair play if you can follow all the technical and mathematical detail here and in the blog post and paper. The plain-English version: this is a big deal. It is a meaningful architectural advance against a major LLM bottleneck. The clearest win is cheaper, more scalable inference, often faster inference too, with a secondary training benefit: reaching the same model quality with less total compute. Excited to see how others apply this in production and build on it. Kudos to @JamesONeil21 for his phenomenal work on this. - **destraynor** (2026-04-09): https://x.com/destraynor/status/2042305356705984935 - Our AI group published a novel finding from pre-training research In short: you can cut KV-cache memory in half by sharing most of the attention structure across heads + keeping small per-head differences—without hurting model quality or speed I'll explain this as best I can https://t.co/BRtsEIQZSy - **fergal_reid** (2026-04-09): https://x.com/fergal_reid/status/2042289448935629239 - We’re excited to share our new form of attention, Low Rank Key Value attention. This is a drop-in replacement to standard MHA that in our tests, reduces KV-cache by ~50%, with even lower test loss, across many scales of experiment. https://t.co/JdQddhehuj - **elithrar** (2026-04-09): https://x.com/elithrar/status/2042276864178168183 - i’m traveling so I missed our latest internal demos… it’s great that I can catch up on things via Jane 😉 https://t.co/cojQYUxwoj - **Padday** (2026-04-09): https://x.com/Padday/status/2042273859496296819 - Hot on the heels of our Apex Model launch (the highest performing Model in the world for CX), and opening access to Apex and our other Models, today we have Fin CLI ! You can now point your AI coding Agent at the Fin CLI to do things like set up Fin on your website or configure your Intercom helpdesk. Let me know what else you'd like Fin CLI to do! Quick 2 min demo of Fin CLI below, check it out! - **martin_casado** (2026-04-09): https://x.com/martin_casado/status/2042266285464309905 - It's only a matter of time before only the model creators have access to the most powerful models. The rest get access to smaller, distilled versions. Or access the models through first party apps and services that don't provide direct access to the token path. The investment needs for training are too high, and distillation too effective to warrant any other future. - **ylecun** (2026-04-09): https://x.com/ylecun/status/2042224846881349741 - @ClementDelangue @guillaumgrallet Mythos drama = BS from self-delusion. - **thdxr** (2026-04-09): https://x.com/thdxr/status/2042038193122312332 - no no i've definitely been doing a lot of work it's all just too dangerous to release - **dexhorthy** (2026-04-08): https://x.com/dexhorthy/status/2041987044017369176 - PLEASE DOCS WRITERS INCLUDE THE IMPORT STATEMENTS https://t.co/4HEsMCokIA - **dexhorthy** (2026-04-08): https://x.com/dexhorthy/status/2041985486085091377 - i am excited to mess with managed agents but crazy to me that The Safety Company™️ is distributing unsigned binaries https://t.co/eRTlwXOXp8 - **simonw** (2026-04-08): https://x.com/simonw/status/2041975451133079719 - The feature I most want from AI labs right now is documentation on which underlying search engines they use when their chat tools run a search OpenAI and Anthropic and Meta AI all have search and I have NO IDEA what index they are using (I would hope tha Gemini uses Google!) - **ankrgyl** (2026-04-08): https://x.com/ankrgyl/status/2041965465891958900 - I feel so vindicated! https://t.co/GaSGxOyDaw https://t.co/ZzNvX79g4K - **darraghcurran** (2026-04-08): https://x.com/darraghcurran/status/2041944297654657319 - Love this - easier and quicker than ever to sign up and install Fin and Intercom in your product/app/business. The future is agentic. https://t.co/yZ0tDscFKx - **brian_scanlan** (2026-04-08): https://x.com/brian_scanlan/status/2041942364051145120 - This is no longer an experiment. Announcing the Fin CLI - the agent-first way of signing up to, configuring and using Fin and the Intercom helpdesk. Ask your agent to use fin dot ai slash cli to set Fin up from scratch. https://t.co/GwH7k6J2jj - **ritakozlov** (2026-04-08): https://x.com/ritakozlov/status/2041935070244602366 - proud to have @replicate be the first provider to support seedance 2.0, a truly incredible video model https://t.co/Spjq4wCXfM - **KentonVarda** (2026-04-08): https://x.com/KentonVarda/status/2041922361499304329 - Cap'n Web APIs are better than REST, especially for AI agents. Here's a talk I gave explaining why: https://t.co/5pD6hhxn7Z In short: - TypeScript types are an excellent way to explain APIs to agents. - Cap'n Web excels at exposing TS APIs over a sandbox boundary. - **nandafyi** (2026-04-08): https://x.com/nandafyi/status/2041913203689713749 - now live! check it out in your CF workers page :) https://t.co/66XHVlDILD https://t.co/xCkrbH5K52 - **BraydenWilmoth** (2026-04-08): https://x.com/BraydenWilmoth/status/2041887708289618049 - interesting https://t.co/y1ZXyw45hX - **JacobTref** (2026-04-08): https://x.com/JacobTref/status/2041876743666200980 - Alzheimer’s is one of medicine's hardest unsolved problems, and one of the most devastating. At the OpenAI Foundation, we believe AI is well suited to its complexity. We're directing over $100M to scientists mapping the disease, designing drugs, & more. I wrote about it here: https://t.co/wOkiE78KUo - **FredKSchott** (2026-04-08): https://x.com/FredKSchott/status/2041668123544220070 - bingo https://t.co/BtSgiEqj3f https://t.co/scLwKQcXO8 - **thsottiaux** (2026-04-07): https://x.com/thsottiaux/status/2041655710346572085 - Three million people are now using Codex weekly - up from two million a little under a month ago. Incredible to see the growth. Thank you to all of you and to the ecosystem we’re part of. To celebrate, we’re resetting rate limits so you can keep building, and we’ll reset them every additional 1M users until we reach 10M, so we can keep celebrating along the way. Enjoy and thank you! - **reed** (2026-04-07): https://x.com/reed/status/2041651704136130986 - @mattparlmer @WillManidis @grok has uber seen reduced ridership in cities where waymo operates? - **thdxr** (2026-04-07): https://x.com/thdxr/status/2041642239365955849 - guys i had a one of a kind idea for a business sandboxes for coding agents - **theo** (2026-04-07): https://x.com/theo/status/2041641012016378004 - I would highly, highly recommend you make sure your phone, computer, browser, and important apps are all updated and on the latest versions. - **garrytan** (2026-04-07): https://x.com/garrytan/status/2041621778787963174 - SFO approved Waymo — then buried it at the Rental Car Center. Meanwhile, Uber and Lyft run 800K trips/month from the main garage. Zero safety rationale. Pure incumbent protection. https://t.co/53a332crtg - **boyney123** (2026-04-07): https://x.com/boyney123/status/2041587296709779600 - Been fucking around with CloudFlare, learning about durable objects, Agent SDK and Dynamic workers. Built this POC, that lets you describe an API you want, dynamic worker builds/hosts it, each user has it's own durable object and APIs + State. Been super fun 🤩 https://t.co/c7scBr5TG0 - **kevinweil** (2026-04-07): https://x.com/kevinweil/status/2041573799775367525 - 💥 New in Prism today: Paper Review, an AI workflow for reviewing technical and scientific papers. This is the opposite of AI slop: we're using AI to improve scientific rigor, correctness, and reproducibility. https://t.co/ngtduaXDqx - **Railway** (2026-04-07): https://x.com/Railway/status/2041512880290296198 - We moved Railway's entire frontend off Next.js. Two PRs, zero downtime. Builds went from 10+ minutes to under two. 200+ routes on @vite_js + @tan_stack Router, instant HMR, and dev server startup in seconds. @vrzgc's full breakdown: https://t.co/01jxBHWEqZ - **threepointone** (2026-04-07): https://x.com/threepointone/status/2041457759284072954 - yes yes a 1000 times yes after the, er, recent drama, I went into a rabbit hole exploring the space, and it's like my brain's been reprogrammed lemme put it this way: there's nothing _fundamental_ about filesystems. or terminal commands. or _even your favourite programming language_. but there _is_ something fundamental about storage. or lambda calculus. the future of computing looks more like math+symbolic execution than unix+whatever's in the training set (and it would be very bad for humanity if this wasn't true!) (separately: theo's so far ahead of other js youtubers. always pleasantly surprised at the depth he goes to.) - **thdxr** (2026-04-06): https://x.com/thdxr/status/2041000051904856088 - i've mentioned this a bunch of times but i see this all as companies having learned from the past whenever there's a major technical shift like this the first wave of companies end up being projected to be huge winners of course the bottom of the stack is the most valuable, the whole world will build on top and pay them! but it usually turns into some race to zero commodification fight with a bunch of other competitors and the layer built on top becomes more valuable because they have the end relationship with the actual customer so this time around i think the companies are seeing if they can do both layers - **openclaw** (2026-04-06): https://x.com/openclaw/status/2040998570317197607 - OpenClaw 2026.4.5 🦞 🎬 Built-in video + music generation 🧠 /dreaming is now real 🔀 Structured task progress ⚡ Better prompt-cache reuse 🌍 Control UI + Docs now speak 12 more languages Anthropic cut us off. GPT-5.4 got better. We moved on. https://t.co/T3LaSJYOvU - **theo** (2026-04-06): https://x.com/theo/status/2040988058427023668 - They’re trying so hard to kill OpenClaw that they made Claude Code unusable. I’m done. I give up. https://t.co/TjqJJY4Mxi - **duncancampbell** (2026-04-06): https://x.com/duncancampbell/status/2040965451606286779 - @__tinygrad__ I have MWs of power available but it’s DC. Would you do a DC-native version so I don’t need to invert just for the racks inside your box to rectify again later. - **fidjissimo** (2026-04-05): https://x.com/fidjissimo/status/2040936371955925081 - These diseases are extremely debilitating; for example MECFS has a higher burden of disease than most diseases, yet is critically underfunded. https://t.co/BMQ4bhdWdm - **fidjissimo** (2026-04-05): https://x.com/fidjissimo/status/2040935739903688956 - If there is one good thing that can come out of having my health issues exposed to the world, it’s raising awareness for complex chronic conditions like POTS, MECFS, Long Covid or EDS. 🧵 https://t.co/nNkKjGAHZm - **dwr** (2026-04-05): https://x.com/dwr/status/2040886739695472717 - Presidio Tunnel Tops is pretty great. https://t.co/lMtOIzKupx - **ankrgyl** (2026-04-05): https://x.com/ankrgyl/status/2040859604037537863 - The fundamental problem with this whole harness craze is that they are designed for a very specific single-player and your choice of insecure or expensive (full access to sandbox) environment. Beyond the cost, this constrains harnesses/agents to a very specific (slow chat or background work). We need to build harnesses that are both efficient and secure, and optimize those for model performance. Additional layers of power that require a sandbox can then be opt-in when needed, not default provisioned for every interaction. Maybe I'm wrong and the correct thing to do is just optimize the F out of the sandbox layer, but that seems unlikely to be efficient at scale. If an LLM's action does not need a sandbox, then managing a file descriptor to stream bytes over the network is outrageously cheaper and more secure than provisioning a virtual machine. I believe that until we build true multi-tenant harnesses, the use cases for agents will continue to be constrained to ~ coding. Which is a lot of stuff! But undersells what agents could do at scale. - **simonw** (2026-04-05): https://x.com/simonw/status/2040846932239851936 - Billing different based on text contained in the system prompt is a really bad look https://t.co/15meK1YIe2 - **ylecun** (2026-04-05): https://x.com/ylecun/status/2040770225990172676 - @elonmusk Thinking in language has limited applications, largely in coding and mathematics where the language itself can help reasoning. But, as I've been saying for years, thinking manipulates mental models in abstract (continuous) representation space. Soooo, xAI gonna use JEPA now? - **signulll** (2026-04-05): https://x.com/signulll/status/2040602383307210892 - the reason why emotional intelligence is extremely rare is simply cuz it is expensive. actually modeling another person’s internal state in real time is computationally brutal.. it requires suppressing your own frame, running a parallel simulation, & updating continuously. the “inference costs” of eq in humans are likely much higher than the most expensive ai model in existence today. - **ylecun** (2026-04-04): https://x.com/ylecun/status/2040542673463824880 - @nxthompson So much BS - **darraghcurran** (2026-04-04): https://x.com/darraghcurran/status/2040468988656791579 - Thanks @jonas - I beg to differ - obviously we've our eyes wide open about the risks here. Are you concerned about the principle of it? i.e. the idea that we would have AI approval in some situations? Or the arbitrary nature of it or aggression of the goal? i.e. chasing this goal will carry high risk of hitting the goal but at the expense of very bad tradeoffs? What would you suggest instead? It's easy to take something out of context and attack it - would love to understand how you'd approach this? - **michellechen** (2026-04-04): https://x.com/michellechen/status/2040455406112108745 - gemma 4 from @googleaidevs now on @CloudflareDev workers ai 💎 https://t.co/OKM8cha3VZ - **dillon_mulroy** (2026-04-04): https://x.com/dillon_mulroy/status/2040453577647206554 - if you want to see a masterclass in product engineering/work, scroll through @MattieTK’s timeline from the past week in launching EmDash he’s taken criticism in stride, engaged with everyone in good faith, iterated on the product, and started building community https://t.co/w4OY136k1e - **MattieTK** (2026-04-04): https://x.com/MattieTK/status/2040447046742626543 - EmDash is at 6.8k GH stars after just three days. 😲 We're blown away by the community gathering on GitHub but it's getting very busy for us! To bring together the hundreds of people enthused by EmDash, we've put together a Discord ⤵️ https://t.co/rfZazbnFVS - **samlambert** (2026-04-04): https://x.com/samlambert/status/2040257771334250557 - AI is in the 90's right now. Some companies are rejecting open source and it will look like the right move for a while. The ones that embrace it in the long run will win. - **boristane** (2026-04-04): https://x.com/boristane/status/2040253876822221118 - based https://t.co/rFzBdf9FLC - **steipete** (2026-04-03): https://x.com/steipete/status/2040209434019082522 - woke up and my mentions are full of these Both me and @davemorin tried to talk sense into Anthropic, best we managed was delaying this for a week. Funny how timings match up, first they copy some popular features into their closed harness, then they lock out open source. https://t.co/Mgmv6YmW2B - **thsottiaux** (2026-04-03): https://x.com/thsottiaux/status/2040185809882742805 - @varunram We're not Apple. We are OpenAI. - **darraghcurran** (2026-04-03): https://x.com/darraghcurran/status/2040121725535322589 - The specialised models we trained to power Fin, out performing the best general models on quality, latency and cost, are now available for others to build their own unique customer service experiences on. And this is just the beginning, we are heavily invested in further improvements. Get in touch and tell us about what you’d like to build. Already running agents at scale, here’s an opportunity to upgrade the performance of your agent. - **ciaran_lee** (2026-04-03): https://x.com/ciaran_lee/status/2040119645210878072 - Excited about this. The timeline is accelerating. https://t.co/gEYnQX1NXV - **destraynor** (2026-04-03): https://x.com/destraynor/status/2040119377639395793 - The #1 question I got last week was "Will @Fin_ai's Apex model be available as an API" The answer, now, is yes. Good read below as to why... - **burcs** (2026-04-03): https://x.com/burcs/status/2040093205601857918 - it's all coming together... https://t.co/GM9oi5jgBN - **dexhorthy** (2026-04-03): https://x.com/dexhorthy/status/2040080664008204390 - I have often said similar - I'm max 1-2 complex things in parallel, and I ALWAYS have a primary task I'm not advancing them both along a workflow together, its "something to work on while primary is blocked / researching / coding / etc" and then as soon as the primary is unblocked, I'm back on it - **dillon_mulroy** (2026-04-03): https://x.com/dillon_mulroy/status/2040078636749836408 - ive been working w/ @elithrar, @mattzcarey, and @thomas_ankcorn on this and it's possibly the most exciting/cool thing ive worked on in my career also finally had the chance to work on a project w/ these 3. some of the most talented and fun colleagues to work with https://t.co/n01Z1UqJjd - **mattzcarey** (2026-04-03): https://x.com/mattzcarey/status/2040076626872996134 - this is the most exciting thing I've worked on since joining and we've done some pretty cool stuff https://t.co/Dhf7x9Yd4r - **destraynor** (2026-04-03): https://x.com/destraynor/status/2040062732905353551 - "If I had asked my customers what they wanted, they would have said a horse powered by an LLM" https://t.co/93F9oM4WiH - **MattieTK** (2026-04-03): https://x.com/MattieTK/status/2040060382627217807 - @hluehrsen @WordPress @Cloudflare I absolutely love this take, it was a huge frustration for me when working with WordPress in the past. The advantage that WordPress has in its history and backwards compatibility is it's biggest weakness. Tear the bandaid off. - **ritakozlov** (2026-04-03): https://x.com/ritakozlov/status/2040050818842345756 - automatic retries for any model landed in @cloudflaredev ai gateway! super useful for those peak US usage moments https://t.co/2GN4ScbWv1 https://t.co/Q315cYntYK - **kysely_** (2026-04-03): https://x.com/kysely_/status/2040031915294573048 - btw i'm in @Cloudflare's emdash too. https://t.co/SFbZzBtvTE - **threepointone** (2026-04-03): https://x.com/threepointone/status/2040018559716278491 - I landed my first 2 prs in the workerd repo: - increased max size of attachments on hibernated websockets from 2kb to 16kb making them a tad more useful https://t.co/PfdUQyxyJW for agents sdk, it made it possible to automatically derive callbacks for authing mcp servers. small thing, but so nice for dx. - persist alarms to disk for local dev https://t.co/EwKdVpj65W this is core to a bigger feature I'm very excited to ship this month. can't wait. - **steipete** (2026-04-03): https://x.com/steipete/status/2039948914112209390 - AM I HOT PEOPLE OR SLOP CANNON https://t.co/HpOLGE6Vih - **scottbuscemi** (2026-04-03): https://x.com/scottbuscemi/status/2039911911459017061 - Evening project: @WordPress 7.0 realtime collaboration over @Cloudflare Workers! This frees up PHP from handling the communications. https://t.co/ZAR9GdRFxb - **eastdakota** (2026-04-03): https://x.com/eastdakota/status/2039902419379794227 - Think this is a fair critique from @photomatt of EmDash. I remain hopeful it’ll bring a broader set of developers into the WordPress ecosystem. https://t.co/7lJ8wnctTd - **flytip** (2026-04-03): https://x.com/flytip/status/2039892200008425931 - Whoop accomplished the vision that Nike Fuelband set out to deliver over a decade ago. Fuelband was visionary, innovative, and it would have set Nike along the path of digital services, data, ai and health+wellness. Instead, Nike prematurely pulled the plug on the program and now Whoop’s market cap is worth 1/6th the market cap of ALL of Nike. Wow. 🤯 - **mattrothenberg** (2026-04-03): https://x.com/mattrothenberg/status/2039875548906733965 - ummm you can create some obnoxiously cool focus rings with the new HTML-in-Canvas API https://t.co/QW6fYewZzp - **willahmed** (2026-04-03): https://x.com/willahmed/status/2039865414755188887 - You have no experience. You’ve never started a company. You’ve never had a full time job. Nike is going to kill you. You’re a kid. You don’t have technical skills. You shouldn’t build hardware. Apple is going to kill you. You can’t build hardware. You can’t measure heart rate non-invasively. Athletes don’t care about recovery. Under Armour is going to kill you. It won’t be accurate. You don’t listen. You’re an ineffective leader. You can’t recruit great talent. You’re going to have to pay every athlete. You can’t measure sleep non-invasively. It’s too expensive to research. Athletes are a small market. The product costs too much to make. The product costs too much to sell. Your valuation is too high. Consumers aren’t going to want it. Hardware is too hard. You should measure steps. Fitbit is going to kill you. You can’t build a marketing engine. You can’t raise enough money. You need a real CEO. Google is going to kill you. You can’t be a subscription. You can’t build a brand. You can’t do consumer in Boston. Your valuation is too high. You shouldn’t make accessories. You shouldn’t make apparel. Lululemon is going to kill you. You can’t predict Covid. Stay in your niche. You are going to run out of money. You can’t build a health platform. Amazon is going to kill you. You can’t measure blood pressure. You can’t get medical approvals. The market is too small. You don’t understand AI. The market is too competitive. It won’t work internationally. The supply chain is too complicated. You can’t build an AI. You can’t raise enough money. It’s too competitive. Healthcare isn’t going to want it. … Just keep going ✌️ - **gakonst** (2026-04-03): https://x.com/gakonst/status/2039856749960974640 - I worry that we've been HackerNews Reply Guy'ing Passkeys. I had this experience again when I had a smol at home hackathon w/ my friends. Passkeys worked on everyone's computer/phone 1 shot. The ones that knew about Passkeys said how much they love them, how much they hate passwords and love FaceID. They also knew about the amazing feature that lets you share a passkey with a friend/device not on the same account. - **MylesUdland** (2026-04-02): https://x.com/MylesUdland/status/2039849432842150103 - @TheStalwart @tbpn @jordihays @johncoogan Many cases in many industries - **TheStalwart** (2026-04-02): https://x.com/TheStalwart/status/2039847391721841087 - An underrated aspect of @tbpn’s success is that @jordihays and @johncoogan actually aspired to make something good and entertaining. Other tech industry-adjacent media often seems kind of bitter. Or like it’s only trying to prove a point. - **MattieTK** (2026-04-02): https://x.com/MattieTK/status/2039840306514808967 - I've seen people saying that @EmDashCMS's UI looks old enough to drink just because the interface is a little simple. I've also seen people praising our "lack of block editor". Funny thing is: this is a 'block editor'. It's just that Gutenberg is so obtrusive, it has haters. https://t.co/WJu0RJ9bln - **MattieTK** (2026-04-02): https://x.com/MattieTK/status/2039835483690148211 - @barckcode I loveeee it! Biggest pain point of your experience? - **zebassembly** (2026-04-02): https://x.com/zebassembly/status/2039834855354126449 - @irvinebroque @dillon_mulroy @kitlangton @msmps_ @nevikashah t shirt size ships ❌ cuisine cook time ships ✅ - **dillon_mulroy** (2026-04-02): https://x.com/dillon_mulroy/status/2039829006623613177 - @zebassembly @kitlangton @msmps_ @nevikashah get me custom spans/traces first 😭 - **zebassembly** (2026-04-02): https://x.com/zebassembly/status/2039823495690125560 - @kitlangton Holy shit this is amazing. @msmps_ @nevikashah what about this for wrangler? - **kitlangton** (2026-04-02): https://x.com/kitlangton/status/2039822258940617132 - I recommend looking into 𝚘𝚙𝚎𝚗𝚝𝚞𝚒 https://t.co/29Cp4MMiYi https://t.co/xJAwAeucDx - **barckcode** (2026-04-02): https://x.com/barckcode/status/2039814940823814654 - 😋 Bueno he estado probando EmDash (El sucesor de WordPress como le ha nombrado Cloudflare) La verdad mola mucho. Que esté basado en Astro además lo hace bastante ligero y eso está muy bien para sitios principalmente estáticos. El admin de momento es bastante simple. No esperéis todo lo que tiene Wordpress pero están en beta. Estoy 100% seguro que si nace comunidad esto se potenciará un montón. En mi caso tenía la landing ya desplegada en Cloudflare y mi blog estaba en otro dominio en Ghost. La verdad que poderlos unificar bajo el mismo sitio y en especial poder gestionar los posts del blog sin hacer un solo cambio en código, directo desde el CMS es una gozada. Tiene algunos bugs aún como que algunas imágenes no renderizan en el Admin, las listas no renderizan en la UI normal y algunas cosillas más pero es normal estando en beta. En mi caso la verdad que cubre lo suficiente como para tirarme a la piscina y ya he unificado la landing + blog bajo este nuevo CMS. Ojo, en realidad aún no lo recomiendo para producción porque en su repo dejan claro que habrá breaking changes en próximas actualizaciones. PERO en mi caso lo hago porque mi sitio apenas sufre cambios, he calcado el mismo diseño que tenía desde principios de 2023 y no creo que cambie en el corto plazo. Además quiero aprovechar para abrir algunas issues al repo oficial de Emdash y me viene bien tener un sitio donde probar los bugs que he ido pillando en el camino. Por último pero no menos importante quiero ver si puedo desde AgentCrew o desde el propio Claude Code conectarlo al MCP/CLI nativo de EmDash para gestionar todo el sitio. Mi objetivo con esto es: escribir un post, pasarselo al agente y que este lo publique con todo con tan solo un mensaje simple en un chat. Podéis ver todo el sitio corriendo en EmDash aquí: https://t.co/wvHCPIoXf5 - **karpathy** (2026-04-02): https://x.com/karpathy/status/2039805659525644595 - LLM Knowledge Bases Something I'm finding very useful recently: using LLMs to build personal knowledge bases for various topics of research interest. In this way, a large fraction of my recent token throughput is going less into manipulating code, and more into manipulating knowledge (stored as markdown and images). The latest LLMs are quite good at it. So: Data ingest: I index source documents (articles, papers, repos, datasets, images, etc.) into a raw/ directory, then I use an LLM to incrementally "compile" a wiki, which is just a collection of .md files in a directory structure. The wiki includes summaries of all the data in raw/, backlinks, and then it categorizes data into concepts, writes articles for them, and links them all. To convert web articles into .md files I like to use the Obsidian Web Clipper extension, and then I also use a hotkey to download all the related images to local so that my LLM can easily reference them. IDE: I use Obsidian as the IDE "frontend" where I can view the raw data, the the compiled wiki, and the derived visualizations. Important to note that the LLM writes and maintains all of the data of the wiki, I rarely touch it directly. I've played with a few Obsidian plugins to render and view data in other ways (e.g. Marp for slides). Q&A: Where things get interesting is that once your wiki is big enough (e.g. mine on some recent research is ~100 articles and ~400K words), you can ask your LLM agent all kinds of complex questions against the wiki, and it will go off, research the answers, etc. I thought I had to reach for fancy RAG, but the LLM has been pretty good about auto-maintaining index files and brief summaries of all the documents and it reads all the important related data fairly easily at this ~small scale. Output: Instead of getting answers in text/terminal, I like to have it render markdown files for me, or slide shows (Marp format), or matplotlib images, all of which I then view again in Obsidian. You can imagine many other visual output formats depending on the query. Often, I end up "filing" the outputs back into the wiki to enhance it for further queries. So my own explorations and queries always "add up" in the knowledge base. Linting: I've run some LLM "health checks" over the wiki to e.g. find inconsistent data, impute missing data (with web searchers), find interesting connections for new article candidates, etc., to incrementally clean up the wiki and enhance its overall data integrity. The LLMs are quite good at suggesting further questions to ask and look into. Extra tools: I find myself developing additional tools to process the data, e.g. I vibe coded a small and naive search engine over the wiki, which I both use directly (in a web ui), but more often I want to hand it off to an LLM via CLI as a tool for larger queries. Further explorations: As the repo grows, the natural desire is to also think about synthetic data generation + finetuning to have your LLM "know" the data in its weights instead of just context windows. TLDR: raw data from a given number of sources is collected, then compiled by an LLM into a .md wiki, then operated on by various CLIs by the LLM to do Q&A and to incrementally enhance the wiki, and all of it viewable in Obsidian. You rarely ever write or edit the wiki manually, it's the domain of the LLM. I think there is room here for an incredible new product instead of a hacky collection of scripts. - **ericclemmons** (2026-04-02): https://x.com/ericclemmons/status/2039795692802093341 - Pages had Deploy Hooks & now Workers do too! Shipped by my teammate @jaeheon_s 🎉 https://t.co/UNyMNRnjvT - **hubermanlab** (2026-04-02): https://x.com/hubermanlab/status/2039790799639175183 - 93 years & 231 days old Ann Esselstyn dead hangs for 2 min 52 seconds! To set new WR. (Maybe we call it live hangs instead now?) Distal strength reflects many things. https://t.co/wpup5t53tj - **Cloudflare** (2026-04-02): https://x.com/Cloudflare/status/2039789898945749389 - One reason @MattieTK joined Cloudflare after roles at places like the Financial Times? Cloudflare’s early bet on x402, the standard for Internet-native payments. Full conversation on @EmDashCMS, the spiritual successor to WordPress, drops next Monday on https://t.co/QfQVicbOi6 https://t.co/wqnSqQr2m6 https://t.co/IqJhJjqn09 - **thdxr** (2026-04-02): https://x.com/thdxr/status/2039787418870583603 - why has no one vibed up really good local otel tooling i want to be able to look at all these traces during local dev - **jeff_weinstein** (2026-04-02): https://x.com/jeff_weinstein/status/2039780886745661928 - congrats to @jordihays, @johncoogan, and the @tbpn team on the acquisition! they built something useful, fun, and uniquely-them quickly, in a world where that increasingly matters. i recall jordi telling me "i'm making a tech tv show", and i was like "sounds cool". and now look. - **dinasaur_404** (2026-04-02): https://x.com/dinasaur_404/status/2039777989487890658 - @yomnashousha @Cloudflare gonna trigger you - **yomnashousha** (2026-04-02): https://x.com/yomnashousha/status/2039777572142072083 - 🚀 Deploy Hooks are now live on @Cloudflare Workers! Trigger a build + deploy from anywhere: your headless CMS, a cron Worker, a Slack slash command, and more! What are you gonna trigger first? 📷👀 https://t.co/CahKHOCKzU - **rohinlohe** (2026-04-02): https://x.com/rohinlohe/status/2039757418390057037 - 25 years ago, Google helped create the first economic model for the web — ads. Today, that model is changing and we (@Cloudflare) are excited to enable every developer and site owner to shape how the world transacts. Honored to be a steward, alongside iconic businesses such as Coinbase, Stripe, Visa, Mastercard, Google, Microsoft, and many more. A special thank you to @programmer and @kleffew94 for their vision, and to Coinbase for their openness to making this an open protocol. Get started today & send your feedback my way: https://t.co/DmyHoJ94b6 - **thdxr** (2026-04-02): https://x.com/thdxr/status/2039750469153607723 - some thoughts on figuring out how to keep producing good work https://t.co/eUCyhVHabE - **jeff_weinstein** (2026-04-02): https://x.com/jeff_weinstein/status/2039735072342028455 - .@stripe is excited to join the newly formed x402 foundation, as a founding member along with @coinbase and @cloudflare, to build an open ecosystem for agent-initiated payments. Businesses on Stripe can accept money via x402 in a few lines of code: https://t.co/bggWrEarjT - **maietta** (2026-04-02): https://x.com/maietta/status/2039733939007627409 - No way! Emdash (Wordpress successor from Cloudflare) literally just imported my client's entire WordPress XML export perfectly, even doing a better job then WP itself when importing. It actually created and assigned the article posts to the correct authors! https://t.co/HyudERLKei - **lessin** (2026-04-02): https://x.com/lessin/status/2039727836559516114 - Waymo + highway makes me WAAYYY more likely to go to SF on any given day then before. hands down. 5x the experience of driving or uber... if only they had a sauna version would almost fully replicate my morning routine working from home. - **astrodotbuild** (2026-04-02): https://x.com/astrodotbuild/status/2039723298984952180 - Please welcome @Mintlify as Astro’s latest sponsor! Thank you Mintlify for supporting our open source mission 💖 https://t.co/Nwh2B65QXu - **grochbegley** (2026-04-02): https://x.com/grochbegley/status/2039693096888213965 - None of my lefty friends like me for this but I am personally pro-Waymo. Driving is incredibly dangerous. We now have many studies that show Waymo’s are genuinely less dangerous. Ride share drivers’ jobs are not more important than saving lives. https://t.co/8nADTLO1L3 - **jessepollak** (2026-04-02): https://x.com/jessepollak/status/2039686562972078464 - cloudflare just launched next gen wordpress with x402 on @base built in https://t.co/LSR3w3ooc8 - **Techmeme** (2026-04-02): https://x.com/Techmeme/status/2039641670879420816 - Cloudflare debuts EmDash, an MIT-licensed, TypeScript-based CMS built on Astro, designed as a serverless "spiritual successor" to WordPress, available on GitHub (Cloudflare) https://t.co/kmJ757I6zp https://t.co/cqU6dh31mf 📥 Send tips! https://t.co/wlNZvXuhJs - **evanyou** (2026-04-01): https://x.com/evanyou/status/2039484405476126794 - 🤯 and it’s built on @vite_js https://t.co/IH8PrFYF3Y - **MattieTK** (2026-04-01): https://x.com/MattieTK/status/2039478208500736360 - We're literally 8 hours into the life of @EmDashCMS being a public project: 1.3k GitHub stars, hundreds of excited posts, dozens of great pieces of feedback, a couple of lovely phone calls, and #1 on hacker news (for a bit). Great day. Thanks so much to everyone trying it out. https://t.co/awhD5FLdaG - **henderson_mark** (2026-04-01): https://x.com/henderson_mark/status/2039443000883917148 - @jbenton *Before reading the story* "Replace WordPress? Ok, buddy. Whatever you say." *After reading the story* "I wanna spin this up TONIGHT!" - **dillon_mulroy** (2026-04-01): https://x.com/dillon_mulroy/status/2039434268263268863 - had a great chat with @kentcdodds yesterday on product engineering - give it a listen (or read!) here https://t.co/ur4VR0B70r - **briancoords** (2026-04-01): https://x.com/briancoords/status/2039414219754656082 - Emdash is fascinating and contains a lot of lessons for WordPress (which they very explicitly chose to target and mimic). Even the name is clever reference to AI and typography. Might need to write some thoughts down! - **boristane** (2026-04-01): https://x.com/boristane/status/2039412158724014083 - stop whatever you're doing and watch one of the greatest teach how to use one of the greatest primitives out there https://t.co/bVmzyWQUof - **mikemcalister** (2026-04-01): https://x.com/mikemcalister/status/2039409898207826391 - For a certain kind of builder, this is cool! Expect to see plenty more of the industry coming after WordPress's marketshare this year. We have work to do! Also kind of funny to create a WP killer but also mimmic its long-dated and exhausted interface. Go big and reinvent it! https://t.co/qdwlf2fLxx - **scottbuscemi** (2026-04-01): https://x.com/scottbuscemi/status/2039408326954668431 - @Cloudflare EmDash + Gutenberg block editor = ...? :) https://t.co/4ZBkIySt9P - **kmelve** (2026-04-01): https://x.com/kmelve/status/2039400693380067397 - Super cool to see Portable Text being used here – which has been battled-tested and open-sourced by @sanity_io for years. https://t.co/nlkvAUOmrt - **aarondfrancis** (2026-04-01): https://x.com/aarondfrancis/status/2039398319114379668 - You've heard of @Cloudflare Durable Objects... but what are they? What do they enable you to build? What are the platform guarantees? What new patterns do they open up? Good news! I just released a completely free series on Durable Objects! https://t.co/7Xsf4nZggt https://t.co/QcdsXnJDGx - **lgladdy** (2026-04-01): https://x.com/lgladdy/status/2039396546601468394 - Interesting! Let's build a plugin and see how it's APIs work. https://t.co/Q7q4nyJ4qH - **johnturner** (2026-04-01): https://x.com/johnturner/status/2039392818184003796 - Having been in the WordPress ecosystem for the last 15 years, this looks very interesting. Incoming video tomorrow with my thoughts. I have in my head what an AI first website builder looks like. WordPress will not disappear over night but the market capabilities are changing very fast. - **dok2001** (2026-04-01): https://x.com/dok2001/status/2039382768807752043 - Really good take by @jdevalk https://t.co/RN9ouPoCnm - **jdevalk** (2026-04-01): https://x.com/jdevalk/status/2039379626837885241 - I recently switched my blog to Astro away from WordPress. The comment I got most is "people still need a CMS". I didn't disagree, but did say this could all be easier with AI. Today, that's proven true, and I was lucky enough to have a sneak peek. #notaprilfools 1/3 - **caelin_sutch** (2026-04-01): https://x.com/caelin_sutch/status/2039378249243934919 - hell yeah https://t.co/d8LIdpcivz - **ritakozlov** (2026-04-01): https://x.com/ritakozlov/status/2039375867051217313 - bet cloudflare launching a wordpress successor wasn't on your 2026 bingo card but! it's wild how much (40%!!!) of the web is still wordpress. we decided it was time for a makeover. so... enter emdash — familiar look and feel, open-source (MIT), built on typescript + astro - **CloudflareDev** (2026-04-01): https://x.com/CloudflareDev/status/2039375772159189336 - Introducing EmDash — the spiritual successor to WordPress. Serverless. TypeScript. Securely sandboxed plugins via Dynamic Workers. https://t.co/AQorxEmiKM - **Cloudflare** (2026-04-01): https://x.com/Cloudflare/status/2039375668232802321 - Introducing EmDash — the spiritual successor to WordPress. https://t.co/WqrG6HQlSB - **MattieTK** (2026-04-01): https://x.com/MattieTK/status/2039375587559588020 - It's my first @Cloudflare blog, and it's a big one. We're rebuilding WordPress as if it were built today. It's end to end TypeScript, works as an Astro plugin, and has secure plugin execution in dynamic workers. It's called EmDash, try it now ⤵️ https://t.co/qe3xXBWnt3 - **dok2001** (2026-04-01): https://x.com/dok2001/status/2039375276346384832 - Every WordPress plugin you install is a blind trust exercise. Full database access. Full filesystem access. No isolation. Hope for the best. EmDash plugins can only do what they declare upfront. Sandboxed. Scoped. Auditable. 96% of WordPress security issues come from plugins. We fixed that. - **dok2001** (2026-04-01): https://x.com/dok2001/status/2039375273758490691 - Announcing: EmDash, the WordPress spiritual successor built for the modern web. TypeScript. Serverless. MIT licensed. x402 for agent-era monetization. MCP server built in. Deploy to Cloudflare or anywhere Node.js runs. Imports your existing WordPress site in minutes. npm create emdash@latest https://t.co/vgqc4tQn30 - **gakonst** (2026-04-01): https://x.com/gakonst/status/2039369923261309112 - We've added 3 advanced feature guides to the MPP docs: 1. Identity & $0 Auth: https://t.co/hbphAc2TXO - this lets you re-autenticate into a service that you've already paid. Great for jobs APIs! 2. Discovery: https://t.co/aFwkeCzgZd - auto-generate openapi.json so that crawlers can find your service 3. Refunds: https://t.co/4GedzRf2pL - how to pay back a user? - **danshipper** (2026-04-01): https://x.com/danshipper/status/2039326982521913406 - if you hide the thinking tokens that generated my AI's response you lower its usefulness to me by 40% - **darraghcurran** (2026-04-01): https://x.com/darraghcurran/status/2039323139872559189 - Nice anecdote about the 2nd order benefits of automation. It creates space for you to focus on the more interesting problems and opportunities. https://t.co/tFu8hyEfw5 - **eastdakota** (2026-04-01): https://x.com/eastdakota/status/2039313039279714625 - We’ll take all the key people in Portugal we can get. https://t.co/2m00gsOloF - **thsottiaux** (2026-04-01): https://x.com/thsottiaux/status/2039248564967424483 - Our Codex dashboards are showing increased rate of users hitting rate limits and since we don't fully understand why I have made the cautious decision of resetting the usage limits for all plans. Enjoy. I also wanted to celebrate us finding a pocket of fraudulent accounts that we banned and have helped us regain some compute. The fight against abuse never stops, but it's important to mark the moment and make it a little shared victory. - **garrytan** (2026-04-01): https://x.com/garrytan/status/2039235671869796399 - It's going to be much much more driven by what you can make and much much less driven by what your resume looks like When one founder can do the work of 100 to 1000 people, all of the old rules have been cancelled. - **garrytan** (2026-04-01): https://x.com/garrytan/status/2039235457280790997 - Guys guys guys, I think you have to look at @steipete and realize The next 10 years are going to look very different than the last 10 years, and you're going to need to write a new pattern match because the old one is broken https://t.co/HPrXKT40Ki - **grinich** (2026-04-01): https://x.com/grinich/status/2039219027705499697 - packed house for All Things Web with @esthor and @andrelandgraf 14 amazing demos from builders, plus and a mountain of pizza 🍕 and we capped it off with AI agent bingo 😂 only in sf! https://t.co/5S0PUSuiq6 - **CSElmendorf** (2026-04-01): https://x.com/CSElmendorf/status/2039207019639607587 - So it looks like CA taxpayers will be shelling out the big bucks to "protect" 160 acres of land on the border of Berkeley and Albany from any housing development. Wouldn't a new waterfront park have more value if more people could live beside it? https://t.co/FhDrWWoxPC https://t.co/793EN2P7wb - **dok2001** (2026-04-01): https://x.com/dok2001/status/2039197289827213444 - https://t.co/vysqCf4Afl https://t.co/0xM1uA3j2u - **brezina** (2026-04-01): https://x.com/brezina/status/2039193480086610305 - The east bay shoreline, dominated by and walled-off by a highway kind sucks. The only way to make it great is to underground that highway. And the only way to find that is through building a new waterfront urban human habitat. It’s windy as hell there, but could be a great place to live and benefit east bay communities currently severed from the bay by highways - **scottbuscemi** (2026-04-01): https://x.com/scottbuscemi/status/2039187989633597898 - @dinasaur_404 I opened an issue in the agents repo about the "Deploy to Cloudflare" dev experience... Would love your thoughts :) https://t.co/fNWjxHfU19 - **soleio** (2026-04-01): https://x.com/soleio/status/2039184876872884479 - There are only four jobs left in tech https://t.co/MDddFTO9zo https://t.co/ei9AWMqCLP - **gakonst** (2026-04-01): https://x.com/gakonst/status/2039133432257933450 - API pricing will look a lot more like ad auctions in an agent-first future. Instead of fixed pricing with tiers, APIs will sell a number of calls per unit of time, agents will bid. It will look like HFT but for agents paying for getting API calls fulfilled faster. https://t.co/uDfZcnnQre - **next_hopself** (2026-03-31): https://x.com/next_hopself/status/2039124219770933307 - Routing has its own section on Cloudflare Radar! What features should fill it up? 🤔 https://t.co/50vBSndznf - **saranormous** (2026-03-31): https://x.com/saranormous/status/2039107773942956215 - https://t.co/xzFDudnHNX - **eastdakota** (2026-03-31): https://x.com/eastdakota/status/2039083495700529556 - — - **michellechen** (2026-03-31): https://x.com/michellechen/status/2039043919346827509 - if i say 'like' one more time ill have to move to LA - **dillon_mulroy** (2026-03-31): https://x.com/dillon_mulroy/status/2039043799989522586 - cool, practical, and timely use case for dynamic workers https://t.co/Y1mRQXGlhP - **michellechen** (2026-03-31): https://x.com/michellechen/status/2039042359577505846 - witness me https://t.co/J4N2dGcgFa - **yrechtman** (2026-03-31): https://x.com/yrechtman/status/2039012253341495462 - https://t.co/CJrgWXvVSx - **michellechen** (2026-03-31): https://x.com/michellechen/status/2039003359680528819 - $ opencode /connect cloudflare workers ai /models @ cf/moonshotai/kimi-k2.5 https://t.co/48DZpQxR96 - **jamesdevonport** (2026-03-31): https://x.com/jamesdevonport/status/2038975397442777514 - @dok2001 Really all I want to do is to pick a Github repo for a project I already have on Vercel and have Cloudflare do whatever has to be done to make it run on Cloudflare. I had to spend quite a while back and forth with Claude Code fixing various things which didn't work. - **darraghcurran** (2026-03-31): https://x.com/darraghcurran/status/2038969478411665826 - A wonderfully detailed technical post about how we approach topic identification/clustering, which is a central part of the Fin system and flywheel that enables our customers to continually improve their automation rate over time. A great example of combining LLMs with traditional ML techniques. - **willahmed** (2026-03-31): https://x.com/willahmed/status/2038946945838776706 - BREAKING: WHOOP RAISES $575M AT $10.1B VALUATION  I am pleased to announce that we’ve raised $575M at a $10.1B valuation to accelerate our mission of unlocking human performance and healthspan globally. This round was led by Collaborative Fund with participation from 2PointZero Group, Qatar Investment Authority (QIA), Mubadala Investment Company, Abbott, Mayo Clinic, Macquarie Capital, Glade Brook, B-Flexion, IVP, Foundry, Accomplice, Affinity Partners, Promus Ventures, and Bullhound Capital alongside a group of individual investors including Cristiano Ronaldo, LeBron James, Rory McIlroy, Virgil van Dijk, and Mathieu van der Poel. This investor group and this moment reflect a powerful evolution underway for Whoop and the broader healthcare market. Whoop was born in performance - trusted by the best athletes in the world to train, recover, and compete at the highest level. That foundation remains core to who we are. You see that in the iconic athlete investors joining this round.  But it also represents our push into broader health.  In the past 12 months, WHOOP has received medical clearances, launched blood testing, and created a platform that has saved lives. Abbott and Mayo Clinic - two of the most respected and influential institutions in global healthcare - are now investors in Whoop. These are organizations that have shaped modern medicine. Their decision to partner with us is a clear validation of where our technology is headed. Healthcare systems around the world are reactive. For too long, they have waited for people to get sick, then intervene. Chronic disease is rising and costs continue to climb. At Whoop, we believe the future looks fundamentally different. We are building the most powerful, personal, preventive health platform in the world - powered by continuous biometric data, advanced analytics, and AI to help people understand their bodies and improve their health in real time. I am grateful to our team, our members, and our partners for believing in this vision. I’ve been building this company for 14 years and I’ve never been more excited for the future. - **avenceslau** (2026-03-31): https://x.com/avenceslau/status/2038924911452791221 - Now on Workflows, if you need to return more then 1MiB from a workflow step, you can, just return a readable stream. https://t.co/i557fDKzfB (Local dev support for readable streams coming soon) - **JackCulpan** (2026-03-31): https://x.com/JackCulpan/status/2038873078667600105 - just spun up https://t.co/sMCjb2tf7K off the back off this tweet + the axios npm package stuff it uses @cloudflare's new dynamic workers to bundle from npm at the edge code runs in an isolated environment with zero network access compare package versions without ever installing - **willccbb** (2026-03-31): https://x.com/willccbb/status/2038818799504478713 - i hope someone at @axios is reporting on this - **romainhuet** (2026-03-30): https://x.com/romainhuet/status/2038677236304245087 - We’ve seen Claude Code users bring in Codex for code review and use GPT-5.4 for more complex tasks, so we thought: why not make that easier? Today we’re open sourcing a plugin for it! You can call Codex from Claude Code with your ChatGPT subscription. We love an open ecosystem! https://t.co/h5I28h1qww - **onusoz** (2026-03-30): https://x.com/onusoz/status/2038567975809106297 - Here is the spec and implementation for this flow. The mermaid diagram includes all the steps I mentioned in the post above, including a shameless AI review ralph loop, and other loops to make CI pass, resolve conflicts and so on I would recommend reading the README and TUNING.md to understand the approach here https://t.co/qCR4zBWPDr - **onusoz** (2026-03-30): https://x.com/onusoz/status/2038565725690900992 - acpx v0.4 ships Agentic Workflows, or as I like to call them "Agentic Graphs" It let's you create node-based workflows on top of ACP (Agent Client Protocol), to drive any coding agent (Codex, Claude Code, pi) through deterministic steps This let's you automate routine, mechanical legwork like triaging incoming PRs, bugs in error reporting, and so on... For example, OpenClaw receives 300~500 new PRs per day. A lot of them are low quality, but they still relate to real issues, so you have to address them somehow You need to: - extract the intent - cluster them based on intent - figure out if the proposed changes are legit, or whether they are slop local solutions, like trying to catch flies instead of drying out the swamp - if the PR is too low quality or the intent is not clear, close them - run AI review on them them and address any issues that come up - refactor them if the changes are half-baked - resolve conflicts - and so on... So that when the PR is presented to the attention of the maintainer, all the routine legwork is done and the only remaining thing is the decision to (a) merge, (b) give feedback to the PR author, or (c) take over the PR work yourself I wanted to build this feature since a couple months now, since Codex got so good. OpenAI models are now good at judging implementation quality, so I found myself repeating the same steps I wrote above over and over I also tried putting all this in a single prompt. But I believe there are workflows that should not be a single prompt, but a sequence of prompts in the same session That is because like humans, LLMs are prone to PRIMING. I claim that putting all steps in the same prompt at the beginning of the context will generally give suboptimal results, compared to revealing the intention to the model step by step Creating such a workflow also gives more OBSERVABILITY into the each step that an agent is supposed to take. Agent generates JSON at the end of each step, and that structured data can be used to monitor thousands of agents running at the same time in an easier way, on a dashboard Similar features have been introduced in e.g. n8n, langflow. But AFAIK they are not integrating ACP like the way I do I wanted to have a fresh approach, and to build an API that I can develop freely the way I want, so I created a new workflow API inside acpx The video is from the workflow run viewer, but that is not where you build the workflow. You build it by using the acpx flow typescript API. See examples/pr-triage in acpx repo Before building that, I started from a Markdown file with a Mermaid chart of the flow I had in mind. The Markdown file acts as a spec for the flow, and I have built the workflow through trial and error. I call this process "workflow tuning" I started working on acpx repo PRs one by one, tuning the flow, slowly scaling to more PRs. Finally, when I felt confident, I ran it in parallel over all external open PRs in the acpx repo. I believe it already saved me hours this week My next goal, if well received, is to set this up on a cloud agent so that it can process the 300~500 PRs the OpenClaw repo receives every day, in real time, as they come in I believe this will save all open source maintainers around the world countless hours and make it much easier to herd and absorb external contributions from everyone! - **michellechen** (2026-03-29): https://x.com/michellechen/status/2038399972257783910 - my entire feed is just https://t.co/zFAGdFYBBn - **credealjunkie** (2026-03-29): https://x.com/credealjunkie/status/2038396078823710952 - If I told you this Mission District site was up for sale in 2014, got entitled for a 117 units project that would not have evicted anyone, didn't get built because, the Mission, got foreclosed on in 2019 and is now up for sale again (7 years later) with an allowable 200-250 units, but still isn't economic to build, would you then understand why rents in San Francisco are rising faster than anywhere else in the US? - **thdxr** (2026-03-29): https://x.com/thdxr/status/2038374776704573786 - we all live in a big ecosystem that all feeds into each other - frontier labs push the bar on what's possible - opensource labs build off of that and make their own discoveries - inference providers make huge capex investments to serve these models - app builders create real $$$ demand to justify continued investment - individual hackers get these models running in all kinds of places - lots of co-designing between these parties that improve things - lots of conflict between these parties that create a good equilibrium no one spot is morally superior to the others - very dumb to get caught up in that line of thinking - **pmarca** (2026-03-29): https://x.com/pmarca/status/2038350516875595924 - Thesis: the problem with AI working in every domain = all the edge cases. Antithesis: domains with lots of edge cases = difficult & time consuming to practically impossible for error-prone people. Synthesis: such domains = where AI agents will do best. (Such as SAAS migration…) - **TysonTrautmann** (2026-03-29): https://x.com/TysonTrautmann/status/2038277981169455228 - This is a good thing for security. Right now attackers have an asymmetric advantage: defenders need to prevent every vulnerability, attackers only need to find one. But defenders see code before attackers. With an oracle that sees all vulns, defenders win because they can run it before shipping. Danger is in the transition window where the oracle is still being rolled out by defenders. - **johnarnold** (2026-03-29): https://x.com/johnarnold/status/2038262919582855428 - Relatedly, it's interesting to think of how different the politics of energy would be. In Prez Bush's 2006 SOTU, he called to "move beyond a petroleum-based economy" and for wind, solar, EVs, ethanol, batteries, zero-emission coal, nuclear, > research $, etc. Different time. 5/5 - **adam_tooze** (2026-03-29): https://x.com/adam_tooze/status/2038232204845691161 - Hard to think of a more important map than this right now, featured on today's Chartbook Top Links: https://t.co/qA2ezld5Aa - **vitrupo** (2026-03-29): https://x.com/vitrupo/status/2038230613912887757 - Jeff Dean says we’re going to have to re-engineer our tools because they were designed for human speed. An AI agent can run 50x faster, but the tools it relies on don’t. So even if the model gets infinitely fast, you only get 2-3x improvement overall. Amdahl’s law still applies. - **signulll** (2026-03-29): https://x.com/signulll/status/2038120040105496987 - starting a consumer company is basically opting into pain as a lifestyle. you need this weird, almost contradictory stack which is taste + timing, future intuition + present execution, & culture fluency + product rigor. most ppl have like… one of these. maybe two if they’re lucky. oh the worst thing is that there’s no clean feedback loop. ain’t no tidy dashboards telling you you’re right. it’s mostly vibes, weak signals, & ridiculously delayed validation if it comes at all (kinda like being in a toxic situationship). you’re effectively betting on something that doesn’t fully exist yet, using instincts you can’t quite articulate, in a market that will happily ignore you until it suddenly doesn’t. this is one of the most asymmetric games you can play but also one of the least coherent while you’re in it. i think of it as playing a video game in hall of fame difficulty with no tutorial & half the UI missing. - **jdcmedlock** (2026-03-29): https://x.com/jdcmedlock/status/2038085490013040677 - The view of Mount Diablo from Briones regional park https://t.co/KB4XihkWM1 - **chintanzalani** (2026-03-28): https://x.com/chintanzalani/status/2038026663867330850 - The only 4 jobs that will remain at tech companies. Credits: @yrechtman https://t.co/GsB7kIHqAt - **ramez** (2026-03-28): https://x.com/ramez/status/2038023017557835979 - End corn ethanol. https://t.co/fHVoo8J116 - **ramez** (2026-03-28): https://x.com/ramez/status/2038002883749359780 - China's cleantech exports are surging. A couple observations: 1. All of this is from before the US invasion of Iran and current oil price spike. We should expect these numbers to keep growing, especially EVs. 2. US tariffs haven't stopped China's cleantech exports from growing. They've just hurt Americans. 3. Batteries and EVs are each now much larger export businesses for China than solar. Solar export volume has grown but prices have dropped so much that solar exports in dollars have dropped quite a bit. Batteries and EVs are not only larger than solar is now - they're each larger than solar ever was. We're in entirely new territory. - **thdxr** (2026-03-28): https://x.com/thdxr/status/2037980443627147443 - he's on his way to a full recovery! https://t.co/9N1WjZITUD https://t.co/o1zdGaxuDb - **Jesse_Livermore** (2026-03-28): https://x.com/Jesse_Livermore/status/2037974155581165854 - Obama looking pretty smart in hindsight. - **romainhuet** (2026-03-28): https://x.com/romainhuet/status/2037973734641062378 - This happens a lot! https://t.co/H4GFaoPlc5 - **thdxr** (2026-03-28): https://x.com/thdxr/status/2037938305723957747 - when people pitch me concepts they're excited about they focus on how it works but i want to understand how a user goes from not caring about it, to being interested, to understanding, to evangelizing 9/10 good ideas don't have an answer for this, which makes them bad ideas - **threepointone** (2026-03-28): https://x.com/threepointone/status/2037826249293279338 - canon event, mark the day. https://t.co/pQwROLD4AU - **mrramibanna** (2026-03-28): https://x.com/mrramibanna/status/2037766403059843578 - We launched Stripe Projects on https://t.co/kinzdTuOPf this week and the response has been incredible. Huge thanks to all the developers and service providers who’ve already tried it, sent thoughtful feedback, shared ideas, and offered to build with us. We’ve heard from a huge number of people already, and we’re working through it all. 🧵: - **_chenglou** (2026-03-28): https://x.com/_chenglou/status/2037715519277760531 - Project’s at https://t.co/meTmVhJXlk `npm`/`bun install @chenglou/pretext` and throw your AI into making cool demos for it! - **_chenglou** (2026-03-28): https://x.com/_chenglou/status/2037713766205608234 - My dear front-end developers (and anyone who’s interested in the future of interfaces): I have crawled through depths of hell to bring you, for the foreseeable years, one of the more important foundational pieces of UI engineering (if not in implementation then certainly at least in concept): Fast, accurate and comprehensive userland text measurement algorithm in pure TypeScript, usable for laying out entire web pages without CSS, bypassing DOM measurements and reflow - **reed** (2026-03-27): https://x.com/reed/status/2037649882203439361 - @agarwal unfortunately it’s very not safe. need more waymo - **wesbos** (2026-03-27): https://x.com/wesbos/status/2037622760860705000 - Made it to SF. I take back everything I haven’t felt the sun in 6 months. A robot just brought me Uber Eats. I’ve never felt more agentic @sentry / @syntaxfm / @ThePrimeagen / @teej_dv event tonight - **agarwal** (2026-03-27): https://x.com/agarwal/status/2037619829734273455 - Just got a bike. It’s kind of amazing how much of SF is faster this way than driving. If only it was safe... https://t.co/xi4uQ6p8uL - **boristane** (2026-03-27): https://x.com/boristane/status/2037590989670629548 - honestly, nobody spends "hours" looking through logs or traces to find the root cause of any issue this is BS that most observability providers and now ai sre startups are pushing "hours" is fucking wild - **CloudflareDev** (2026-03-27): https://x.com/CloudflareDev/status/2037589447785501002 - Support for your package manager of choice across all of our dev docs 💅 https://t.co/7UpDCgh2co - **brian_scanlan** (2026-03-27): https://x.com/brian_scanlan/status/2037588797643173890 - Just like everybody else this week, we've also been experimenting with cli interfaces. You can now sign-up for Intercom, configure and install Fin and the messenger from inside your agent. npm install -g @intercom/cli It's wild how fast things are changing. - **momito** (2026-03-27): https://x.com/momito/status/2037588230275498173 - a recent ship to our dev docs: package manager component https://t.co/gz6sFr8TT2 - **reach_vb** (2026-03-27): https://x.com/reach_vb/status/2037557578541007155 - @petergostev it's experimental :)) - **akaphill** (2026-03-27): https://x.com/akaphill/status/2037547531014746169 - In a world where many of us aren't reading the code agents produce, finding new ways to understand the shape of what you're building is more important than ever. https://t.co/pExU9S3Ccu - **darioperkins** (2026-03-27): https://x.com/darioperkins/status/2037448274689274346 - thanks to everyone who shared this. For those asking about the main differences: - credit spreads had blown out before the 1990 oil shock. The credit cycle had already turned - the corporate sector was in deficit in 1990, whereas that is not the case today (although position today somewhat flattered by superstars) - the US was a net energy importer in 1990 and the oil intensity of GDP was higher - the dollar was falling, in part because the RoW (esp Europe) was expected to absorb the shock better. Germany was about to get its reunification boom - **dok2001** (2026-03-27): https://x.com/dok2001/status/2037381585503211584 - @grantkiely @irvinebroque happy path help needed - **thsottiaux** (2026-03-27): https://x.com/thsottiaux/status/2037346989244096581 - Hello. We have reset Codex usage limits across all plans to let everyone experiment with the magnificent plugins we just launched, and because it had been a while! You can just build unlimited things with Codex. Have fun! - **samlambert** (2026-03-27): https://x.com/samlambert/status/2037342912137368021 - i fucking love california - **dkundel** (2026-03-27): https://x.com/dkundel/status/2037340173877620789 - Plugins are incredibly helpful 💖 One example that goes beyond traditional coding: I created the first draft of my slides for tonight's @resend meetup in Google Slides directly from Codex with plugins 🙂 It used our corporate slide template to structure the slides. https://t.co/n6wwQsvUGQ - **kristianfreeman** (2026-03-27): https://x.com/kristianfreeman/status/2037339689477190132 - Slowly building a collection of tools to manage my personal life and work on Workers and Agents SDK. They just work and rarely need updates. Skill file gets ingested into my larger agent system. Having lots of fun - **CloudflareDev** (2026-03-27): https://x.com/CloudflareDev/status/2037336582815175122 - Cloudflare Plugin (Skills + Code Mode MCP) 🤝 Codex Give Codex the full power of the Cloudflare API via MCP and best practices via a set of curated Agent Skills One click to add in Codex! https://t.co/i23snbq21b https://t.co/9NFBnlKD6i - **reach_vb** (2026-03-27): https://x.com/reach_vb/status/2037334254619709755 - all of the plugins released today are open source - enjoy! https://t.co/0F0xI4Xeyt - **OpenAIDevs** (2026-03-26): https://x.com/OpenAIDevs/status/2037296316104282119 - We're rolling out plugins in Codex. Codex now works seamlessly out of the box with the most important tools builders already use, like @SlackHQ, @Figma, @NotionHQ, @gmail, and more. https://t.co/PQDsLqHGA6 https://t.co/TIbsIUAf6S - **LakeAustinBlvd** (2026-03-26): https://x.com/LakeAustinBlvd/status/2037283914952188091 - Send everything you want to see in Cloudflare’s docs to Mohamed! The speed with which he’s making them better each day is kinda nuts… https://t.co/J1zaG9BGYF - **threepointone** (2026-03-26): https://x.com/threepointone/status/2037267072724455763 - landed a couple of new examples in the agents repo - structured inputs: how can your agent "ask" you questions that you answer as multiple choice/yes or no/ free form text/number slider/ whatever. tldr - use client side tools! - push notifications: how can your agent buzz you with a web push notification, even if you've closed your tab or whatever. - this in addition to the new elevenlabs starter earlier today. npm i agents - **MikeNomitch** (2026-03-26): https://x.com/MikeNomitch/status/2037225370278387719 - Securely connect your container to a Worker via HTTP. Big feature added to Cloudflare Containers that opens up a lot of possibilities... (1/n) https://t.co/j3GKCtjKwX - **wesbos** (2026-03-26): https://x.com/wesbos/status/2037210147886408058 - THANKS EVERYONE. Realtime via PartyServer + Durable Objects. It's almost TOO realtime https://t.co/PyF4iSFerd - **fergal_reid** (2026-03-26): https://x.com/fergal_reid/status/2037203684459716719 - https://t.co/niV386N65m - **ciaran_lee** (2026-03-26): https://x.com/ciaran_lee/status/2037200546084307281 - This is so cool. I think also objectively interesting at the industry level. https://t.co/tTnisBGFUW - **michellechen** (2026-03-26): https://x.com/michellechen/status/2037199961914323428 - a lot of people wondering what we've been up to after @replicate joined @Cloudflare . excited to share one of the major releases we've been working on together — cog 0.17.0 cog is an open-source tool to package up machine learning models into containers. it abstracts away the hard parts like CUDA dependency hell, optimizes model environments, and automatically creates an HTTP server that's ready to serve inference. you can deploy a cog container anywhere a Docker container runs, which includes your own k8s cluster, replicate, and soon.... workers ai. for now, enjoy this piece of the puzzle and a wonderful engineering story. we have lots more that we're working on and excited to bring to life! https://t.co/ZSiET9LXzo - **darraghcurran** (2026-03-26): https://x.com/darraghcurran/status/2037199299394543957 - Big drop.... our new Fin Apex model outperforms all leading LLMs for the customer service use case, the last domino to fall as we build a complete suite of proprietary models powering Fin. https://t.co/bOyekCthOz - **brian_scanlan** (2026-03-26): https://x.com/brian_scanlan/status/2037199229869871487 - As of last week, Fin (in English over chat and email) is using a custom model that Intercom built. At our scale and resolution rate, that's pretty impressive. https://t.co/3VTobt6GP2 - **darioperkins** (2026-03-26): https://x.com/darioperkins/status/2037148335107141994 - 4) The second lesson is about how oil prices interacted with existing vulnerabilities, tipping the economy into recession. (A pattern observed throughout history, with oil-price spikes marking the top of most cycles.) The labour market was shaky. Strains had appeared in credit, but the situation didn't look disorderly. Note: credit had boomed in the late-80s, particularly in leveraged buyouts. The scale of that boom – 8%pts of GDP – is similar in magnitude to the recent run-up in private credit. But it wasn't "systemic". This was no 2008. (In fact, the 1990 recession was one of the most vanilla in history.) - **darioperkins** (2026-03-26): https://x.com/darioperkins/status/2037148324969562566 - 🧵Let me take you back to August 1990. Three weeks ago, Iraq invaded Kuwait, oil prices have surged, and the FOMC is meeting to decide how to respond. The economy looks wobbly. Payrolls just recorded a small decline. Greenspan talks about a credit bubble that has started to deflate. There's more than a few credit cockroaches. But nobody thinks the US economy is sliding into recession. The Maestro urges stoicism. Nobody knows what’s going to happen in the Middle East, central banks cant really alter the outcome, so its best to provide stability – by doing nothing. - **nicoisonx** (2026-03-26): https://x.com/nicoisonx/status/2037129668915016005 - I just ran a test of @CloudflareDev workers observability mcp vs codemode mcp to check a few different logs all at once in Cursor. Obs mcp -> 2 full context windows with a compaction, ~8mins. Perfect answer. Codemode mcp -> ~50k tokens total, 1.5 mins. Perfect answer. that is flippin awesome @mattzcarey - **elithrar** (2026-03-26): https://x.com/elithrar/status/2037108599281873359 - the Discord team showed us this last week! one of the best parts was how every app was behind Cloudflare Access by default. didn’t have to worry about auth, SSO or exposing apps to the outside world by accident 😎 https://t.co/FCSgjom59v - **thenanyu** (2026-03-26): https://x.com/thenanyu/status/2037042617213481410 - I really wish we could reclaim design back to meaning “a plan for arranging elements to accomplish a particular purpose” We got overly obsessed with arranging elements and forgot about the purpose. https://t.co/Ii2EXoeC0J - **FranciscoHPro** (2026-03-26): https://x.com/FranciscoHPro/status/2036962367515570615 - @irvinebroque As I read new cloudflare announcements, so my agent auto-PR it for me https://t.co/w7PbkQ8RpD - **michellechen** (2026-03-25): https://x.com/michellechen/status/2036938836115177646 - btw… we’re hiring a ton send blogs / research / github https://t.co/Htasv0kmrA - **thdxr** (2026-03-25): https://x.com/thdxr/status/2036896592007926258 - my dog got out and got hit by a car today both front legs broken - they're checking to make sure it's nothing else and hopefully he'll fully recover hug your puppies! - **dinasaur_404** (2026-03-25): https://x.com/dinasaur_404/status/2036865001160384541 - You can now set up Dynamic Workers directly from the dash! Go to Worker -> Add Binding -> Dynamic Workers 🙂 https://t.co/ghz03jvNln - **southpolesteve** (2026-03-25): https://x.com/southpolesteve/status/2036853203610837459 - @wesbos I'm going to tell my kids this is Module Federation - **wesbos** (2026-03-25): https://x.com/wesbos/status/2036849234427707749 - Trying Cloudflare's new Dynamic Workers Running code I don't trust in a dynamic worker to produce a UI I don't trust displayed in an Iframe https://t.co/fjjDiDKpvk - **sahinolut** (2026-03-25): https://x.com/sahinolut/status/2036834616037060750 - @gm_mertd @Cloudflare @irvinebroque - **MikeNomitch** (2026-03-25): https://x.com/MikeNomitch/status/2036834595665334405 - Docker Hub images, public and private, are now supported on Cloudflare Containers You can always use the private registry that's automatically added to your CF account, but no need to push there on day one. https://t.co/icXlPS8F92 https://t.co/nCxcHIF1lN - **eastdakota** (2026-03-25): https://x.com/eastdakota/status/2036827179150168182 - This is Google’s DeepSeek. So much more room to optimize AI inference for speed, memory usage, power consumption, and multi-tenant utilization. Lots of teams at @Cloudflare focused on these areas. #staytuned https://t.co/hHoY4sLT2I - **eastdakota** (2026-03-25): https://x.com/eastdakota/status/2036821615279997247 - Big, compelling ruling out of the U.S. Supreme Court today that clearly states networks DO NOT face liability just for carrying content. https://t.co/0VB6r0gWLV - **thenanyu** (2026-03-25): https://x.com/thenanyu/status/2036816839553004024 - Put more bluntly, many designers I've worked with care more about the opinions of other designers than of their users. If you had to choose 1. designers think you're cool, but low usage 2. heavy usage, other designers think its lame which one do you choose? every PM picks 2 https://t.co/leQLSfUdnP - **thdxr** (2026-03-25): https://x.com/thdxr/status/2036802899913699435 - .well-known so hot right now - **dillon_mulroy** (2026-03-25): https://x.com/dillon_mulroy/status/2036786725150744955 - ive been saying this - starting a new greenfield project, one with the intent of being a durable long term project, is WAY harder w/ clankers than working in a decent existing code base https://t.co/bDM7v7QPNK - **brian_scanlan** (2026-03-25): https://x.com/brian_scanlan/status/2036781336732500263 - Writing production software, understanding how it is used and fixing it when it breaks are all intimately connected. I am sceptical of the value of agents that are built to do these jobs in isolation. “AI SRE” is probably solving too narrow a problem. - **CFchangelog** (2026-03-25): https://x.com/CFchangelog/status/2036754091167064340 - Declare required secrets in wrangler.jsonc or toml. Wrangler now validates them locally and for type generation. https://t.co/IJpAJS1W9Y - **brian_scanlan** (2026-03-25): https://x.com/brian_scanlan/status/2036715236649058316 - I’m going to do a deep-dive into Intercom’s use of Claude Code with an appropriately Ruby on Rails flavour at the SF Ruby Meetup on April 14th hosted in Intercom’s office! - **fayazara** (2026-03-24): https://x.com/fayazara/status/2036536130330566855 - Here's me spinning up multiple dynamic workers on the fly in milliseconds. they're all isolated btw https://t.co/PFL8olhhcD https://t.co/rIKgBcDSA9 - **0xblacklight** (2026-03-24): https://x.com/0xblacklight/status/2036534699582255329 - lots of folks running expensive sandboxes but really all you need is a filesystem but really you don't even need a filesystem, you just need a filesystem API that frontends something like a database (often you care a lot about ACID compliance or indexability etc; @jeffreyhuber talks about this) you can do this with FUSE and lots of people are building really cool things this way but really you don't even need a filesystem API, because agents don't see the POSIX APIs they just see tokens in and tokens so really all you need is something that looks like a file system but frontends whatever you want - S3, Postgres, Chroma, durable streams, whatever coding agents are great "everything agents" but the problem is that you use an off-the-shelf harness it marries you to the filesystem so you get stuck with FUSE or NFS hacks OR you have to inject a bunch of extra MCP tools that are parallel to the 'real' fs tools but if you own the harness, you can own the control flow this lets you separate the tool INTERFACE from the tool EXECUTION. your tool can look like a normal FS read tool to the agent, but you can use whatever backend you want for the execution logic this unlocks lots of exciting things but it requires you build your own harness or use something more customizable - **soleio** (2026-03-24): https://x.com/soleio/status/2036524585685045758 - This feels like the longest, densest Q1 of all time. - **soleio** (2026-03-24): https://x.com/soleio/status/2036516255260414167 - When the economic logic for every incremental PM or eng hire grows clearer by the day, designers must understand how they’re being evaluated on a relative basis. Learning new tools and methods is not optional. The nature of your work will fundamentally shift with agentic tech. https://t.co/lH7fu0geNK - **southpolesteve** (2026-03-24): https://x.com/southpolesteve/status/2036515967295979843 - Cloudflare Dynamic Workers in open beta today! Built a fun demo: @vite_js + Dynamic Workers where every file save becomes a new worker. Hot reload, but for production. Just another vite plugin. Code: https://t.co/JLpMC1MxXB https://t.co/Zc5vkbZ9JQ - **soleio** (2026-03-24): https://x.com/soleio/status/2036512905785659508 - Wake up, designers. While jobs across other functions are surging through the AI transition, design roles are stagnant. Don’t take refuge in craft and taste. All members of technical staff must demonstrate newfound productivity. If that’s a topic you’ve avoided because it involves shipping or knowing your business model inside-out, you will struggle to make the case for your own field. When designers can’t articulate their company strategy, it’s usually because they are neither its authors nor its executors. Seeking new leverage is no longer optional. - **sahinolut** (2026-03-24): https://x.com/sahinolut/status/2036475556942979141 - dynamic workers basically framemogged entire market of firecracker wrappers - **RhysSullivan** (2026-03-24): https://x.com/RhysSullivan/status/2036469939205493234 - incredibly excited to build ontop of dynamic worker loaders the coolest part to me is being able to dynamically load npm packages, so much potential there https://t.co/6OqRH7kWGd https://t.co/qCMQipQm35 - **dinasaur_404** (2026-03-24): https://x.com/dinasaur_404/status/2036463110744138153 - Dynamic Workers are the future!! 🔥 "they are literally so easy to use" — me, in this video, showing you how to use them https://t.co/vGfwWshl4j https://t.co/gbXYQ1IiXK - **LakeAustinBlvd** (2026-03-24): https://x.com/LakeAustinBlvd/status/2036462079368954100 - At this point I am pretty sure the fastest way (for now) that you can improve AI/agent tooling and automation for your teams outside of R&D is to: 1) Create an email alias where anyone can send jobs they don't want to do with the expectation they will get a response/solution. 2) Staff that with your most AI-agent-pilled lunatic, or a couple of them, and have them manually operate it. Let them filter out the silly requests. Their mission is to find patterns and never repeat the same manual step twice. 3) They iterate on the skill files, workflows, and pipelines that is needed to make it not manual. Because doing it manually sucks. And those materials get better every cycle. 4) Package up and release that tooling so the use cases can be self-serve outside of the email alias. 5) Repeat. - **southpolesteve** (2026-03-24): https://x.com/southpolesteve/status/2036452002318692380 - vinext is one month old today! It's a Vite plugin that reimplements the Next.js API surface. Your existing Next.js app, running on Vite. Deploy to Cloudflare Workers, Vercel, Netlify, AWS, or any Node.js server. It's just a Vite plugin, so `vite start` is all you need to ship to production. One of the biggest questions I got was "how will you maintain this?" The project was always about pushing the limits of what AI can do, and maintaining it has been no different. We use AI to triage, fix issues, review, merge PRs. The numbers over the last month: 395 merged PRs. 52 contributors. 34 releases in 27 days. It's been awesome to see a community form around this. Huge thanks to @James_Elicx, @Divkix, @hyoban_cc, @jokull, @Dayifour, @yunus25jmi, and everyone else who jumped in and shipped real code https://t.co/zGvTui0HGN - **Zackary_Chapple** (2026-03-24): https://x.com/Zackary_Chapple/status/2036442657744957728 - Ok, this is actually worth shouting from the mountain tops about. SO MANY EXCITING THINGS https://t.co/N4sfLz2aE7 - **yusukebe** (2026-03-24): https://x.com/yusukebe/status/2036442428719374567 - You can run Hono app in Dynamic Workers with @cloudflare/worker-bundler!! Super cool!! https://t.co/9epiZ36OHv https://t.co/whkXaP7qr7 - **BraydenWilmoth** (2026-03-24): https://x.com/BraydenWilmoth/status/2036440409027604899 - Dynamic Workers are lightweight secure sandboxes. Been using them for various my projects and they are incredibly powerful. 100x faster and 1/10th the memory. https://t.co/yOUgQRN6AT - **threepointone** (2026-03-24): https://x.com/threepointone/status/2036436480990614006 - new toys: - codemode: chew through multiple tool calls in one/few shots - shell: local fs, that you can back with do/d1/any sql db(!) and r2. git support incoming. - worker-bundler: package npm deps and serve full apps (!) do you see what's happening. https://t.co/3IbEcofRdr - **eastdakota** (2026-03-24): https://x.com/eastdakota/status/2036435478648594644 - This is a big deal. https://t.co/G8PE45EH7g - **mattzcarey** (2026-03-24): https://x.com/mattzcarey/status/2036435304911978949 - sandboxes just got 100x faster https://t.co/oXoOYKPaeD - **burcs** (2026-03-24): https://x.com/burcs/status/2036435172623622571 - classic cloudflare... changing the current paradigm on containers/sandboxes by reinventing them to be lightweight, less expensive, and ridiculously fast https://t.co/d3CJFUQAks - **KentonVarda** (2026-03-24): https://x.com/KentonVarda/status/2036434396035666158 - By the way, I gave a talk a few months ago on the future of agents writing code and why Dynamic Workers are ideal for this use case. If you like to consume information in video form, check it out! https://t.co/dzJuzBIA3J - **dillon_mulroy** (2026-03-24): https://x.com/dillon_mulroy/status/2036434272907661643 - the best sandboxing technology is now in open beta, go try it out and let us know what you think! https://t.co/P7bEpheoOq - **elithrar** (2026-03-24): https://x.com/elithrar/status/2036432594896642462 - sandboxes, but make them incredibly fast, resource-light, and truly ephemeral. make no mistakes. https://t.co/OwJgeAVEdE - **KentonVarda** (2026-03-24): https://x.com/KentonVarda/status/2036431085450433017 - This is all made possible by the isolation model that Cloudflare Workers has been built on since the beginning: isolates, not containers. It's kinda crazy how the thing we've been doing differently and perfecting for 9 years is suddenly so relevant to AI. - **kentcdodds** (2026-03-24): https://x.com/kentcdodds/status/2036430932727476453 - This is a very cool feature. I've been using it for a week to power my own AI assistant with CodeMode https://t.co/94lXbgpHbl - **dinasaur_404** (2026-03-24): https://x.com/dinasaur_404/status/2036430761000108088 - Dynamic Workers are in open beta! Think of them as Workers on-demand — starts in ms, infinitely scalable, and so easy to use 🙂 Perfect for agents running code, Code Mode, & custom automations Plus 3 new helper libraries, documentation, & examples to help you get started👇 https://t.co/lld1GbmL3T - **ritakozlov** (2026-03-24): https://x.com/ritakozlov/status/2036430145460883851 - dynamic workers are now in open beta, and available for anyone to try! 🚀 crazy that the isolates bet we made 9 years ago is the perfect fit for the current era of giving agents dynamic execution environments https://t.co/2CjDuS8cYy - **KentonVarda** (2026-03-24): https://x.com/KentonVarda/status/2036430052238254301 - Dynamic Workers are now in Open Beta, all paid Workers users have access. Secure sandboxes that start ~100x faster than a container and use 1/10 the memory, so you can start one up on-demand to handle one AI chat message and then throw it away. Agents should interact with the world by writing code, not tool calls. This makes that possible at "consumer scale", where millions of end users each have their own agent writing code. https://t.co/i6Gz8xz9BZ - **KentonVarda** (2026-03-24): https://x.com/KentonVarda/status/2036430054205366661 - But that's only half of it. Lots of people today are talking about keeping API keys out of your agent sandbox by injecting them into requests on the way out. You can do that with Dynamic Workers, but you can also do *much better*. Dynamic Workers let you block internet access entirely, and then constructively grant the sandbox access to specific TypeScript interfaces, via Cap'n Web RPC. Essentially you can invent your own "Workers bindings" to give it, pointing at whatever you want. A chat room. A git repo. Etc. You can define the RPC interface to support exactly the operations you want to support, so it's totally obvious what the agent can and can't do. Then you just tell the agent the TypeScript type definition of the interface and let it code. This is *much* cleaner than intercepting HTTP requests, and easier for the agent to understand, too. Check out the docs here: https://t.co/ZiPdtWXZqj - **Cloudflare** (2026-03-24): https://x.com/Cloudflare/status/2036429081214886183 - We’re introducing Dynamic Workers, which allow you to execute AI-generated code in secure, lightweight isolates. This approach is 100 times faster than traditional containers. https://t.co/c36Vkb7I0R - **dillon_mulroy** (2026-03-24): https://x.com/dillon_mulroy/status/2036232419330674880 - i wonder if you could run your vitest test suite in parallel using dynamic workers - 1 test per worker and massively parallelize testing - **samlambert** (2026-03-23): https://x.com/samlambert/status/2036188043221025041 - should we open source pg_strict? - **alantomusiak** (2026-03-23): https://x.com/alantomusiak/status/2036138395534229738 - Overall, I was excited by how tractable of a problem this seemed. Blocking the effect of a bacterial toxin on the genome is solvable by both pharmacological and probiotic approaches. The end result would be a 20% drop in the second most deadly type of cancer! https://t.co/Uamn0FS5Yi - **alantomusiak** (2026-03-23): https://x.com/alantomusiak/status/2036138383584665964 - After a long hiatus, I'm back to writing - starting with a very topical piece. Why are colorectal cancer rates spiking in the young? The usual answers are true: changes in diet, exercise, environment. But there's also a much more direct player involved https://t.co/c0FCX8slIY - **southpolesteve** (2026-03-23): https://x.com/southpolesteve/status/2036129527034786217 - Its wild to me that grep/ripgrep is state of the art locally for agents. The harnesses should ship semantic local search and indexing - **mwylde** (2026-03-23): https://x.com/mwylde/status/2036119867254882427 - We just shipped a huge release for R2 SQL, the @CloudflareDev query engine for Apache Iceberg tables stored in R2 . The short version: much more of the SQL you'd expect to work, now does. CTEs, case statements, aliasing, complex types, 100s of new scalar and aggregate functions https://t.co/8uWp4I3bmD - **samlambert** (2026-03-23): https://x.com/samlambert/status/2036111784524853589 - Traffic Control is the database feature I've wanted for a decade at least. https://t.co/j4m3NEg2br - **MarcSelwan** (2026-03-23): https://x.com/MarcSelwan/status/2036101940371956035 - So what are some of the fun things you can do now? How about an AI model recommendation query for specific agent tasks (IE coding, docs writing, search, etc)? I've been ingesting telemetry captured by my own instance of Cloudflare's AI Gateway where I've been running several agents across different CF-hosted models. This query runs across millions of data points and returns in a few seconds, making it suitable for near-realtime adjustments in agent logic. More in-depth post on this coming soon™️. **Note - this is meant to be illustrative of what you can do and not scientific or optimized - **MarcSelwan** (2026-03-23): https://x.com/MarcSelwan/status/2036101937087889485 - The R2 SQL team at @CloudflareDev is excited to release our biggest update yet with 190+ SQL functions 👉 https://t.co/qb0g6drI0C For those who don't know, R2 SQL is Cloudflare's new distributed SQL engine for querying Apache Iceberg tables stored in R2 Data Catalog. We've now unlocked most of DataFusion along with some help from https://t.co/I9Cs0fiDnO 🦀 Take a look at the SQL reference docs to see all the fun new functions to play with! https://t.co/YTvufDcWKx And don't forget to celebrate with your friendly neighborhood R2 SQL engineer @jeromeschneider @jonathanc_n @throwaway24522 and Sergio Esteves (not on X). - **ankrgyl** (2026-03-23): https://x.com/ankrgyl/status/2035918724587573527 - agent traces are about 2 orders of magnitude larger than traditional traces - **danshipper** (2026-03-22): https://x.com/danshipper/status/2035842017553465814 - new model for engineering team structure in 2026: 2 people only one pirate and one architect the pirate's job is to move as fast as possible to develop valuable, shipped product features by vibe coding. the architect's job is to turn the product surface discovered by the pirate into a reliable, structured machine—also by vibe coding, but at a slower, more well-reasoned pace. every product needs a pirate but most product's only need an architect once they some form of PMF, and in that case they usually don't need one full-time. architects can work across many codebases and solve interesting technical challenges. pirates go hard on a product that they own end-to-end. - **michellechen** (2026-03-22): https://x.com/michellechen/status/2035785311318380615 - can you believe llama4 is not even a year old yet.. feels like decades ago - **Brad_Setser** (2026-03-22): https://x.com/Brad_Setser/status/2035712235075424298 - But it is more accurate -- Before the latest oil shock the really big surpluses were China, Korea, Taiwan and Singapore (more an investment income dollar but ... ) https://t.co/lc59NB6akb - **Brad_Setser** (2026-03-22): https://x.com/Brad_Setser/status/2035697455153442939 - Petrodollars! Nothing produces more heated discussion and, in my experience, less insight. Myths trump facts, because the actual data is a bit obscure -- But here is the most important thing to know. Before the Hormuz crisis, the flow of petrodollars had more or less dried up 1/many - **jdcmedlock** (2026-03-22): https://x.com/jdcmedlock/status/2035538132935053340 - Chamath says blue states "will eventually lose so much tax revenue as to be effectively insolvent." But blue state revenue has been growing faster than red state revenue for the last decade, because most economic growth happens in blue states. Impact of migration is modest. https://t.co/dAV3gYES2C https://t.co/0UUqbyNM92 - **dillon_mulroy** (2026-03-21): https://x.com/dillon_mulroy/status/2035466364685553918 - funny how we’ve spent the past decade optimizing the hell out of ssr, ppr, isr, ssg, etc just for none of it to matter for agents and the future of product consumption - **parmita** (2026-03-21): https://x.com/parmita/status/2035447780555255822 - I just asked myself the most important question I’ve ever asked. What if, god forbid, I had cancer right now? How would I save my life and would I be able to do it without Precigenetics? The answer made me cry. Here’s EXACTLY how I would save my own life TODAY. 🧵 - **Vercantez** (2026-03-21): https://x.com/Vercantez/status/2035431659290730914 - pi + ghostty running entirely in a cloudflare workers durable object. sqlite based file system + js code exec + cron support. The best part is it can deploy worker sites using Dynamic Worker Loaders. https://t.co/CflJjXjchP - **kentcdodds** (2026-03-20): https://x.com/kentcdodds/status/2035142182785753412 - General feedback for @CloudflareDev is to make more UI state routed. Whether that be API tokens or a domain search, I want to be able to navigate forward/backward/deep link with the URL. - **gakonst** (2026-03-20): https://x.com/gakonst/status/2035122435343933950 - https://t.co/VVgeSRzzr6 - **thomasgauvin** (2026-03-20): https://x.com/thomasgauvin/status/2035028244877996240 - Workers VPC Services now have built-in observability 📊 each service gets a Metrics tab — connections, latency, and errors over time. errors are broken down by who's responsible: your origin, your config, or Cloudflare + full error code reference in the docs so you know exactly what went wrong and how to fix it 🔍 - **tibotiber** (2026-03-20): https://x.com/tibotiber/status/2034939448258121954 - Hey @CloudflareDev, just wanted to say whoever works on the redesign of the dashboard is doing an amazing job. It's such a delight to use now 🧡. - **dkundel** (2026-03-20): https://x.com/dkundel/status/2034826507953287500 - We love the open ecosystem ❤️ use your ChatGPT subscription with Codex, OpenCode, Pi and many more ❤️ https://t.co/zWMQtHkuFb - **brian_lovin** (2026-03-20): https://x.com/brian_lovin/status/2034791987325669812 - Everyone makes fun of Eight Sleep until it's 84º in SF and they don't have a liquid-cooled mattress. Suckers. - **SIGKITTEN** (2026-03-19): https://x.com/SIGKITTEN/status/2034774718419476811 - MEGA CODEX https://t.co/nIU1sD0DNs - **thdxr** (2026-03-19): https://x.com/thdxr/status/2034730036759339100 - opencode 1.3.0 will no longer autoload the claude max plugin we did our best to convince anthropic to support developer choice but they sent lawyers it's your right to access services however you wish but it is also their right to block whoever they want we can't maintain an official plugin so it's been removed from github and marked deprecated on npm appreciate our partners at openai, github and gitlab who are going the other direction and supporting developer freedom - **michellechen** (2026-03-19): https://x.com/michellechen/status/2034722322855153888 - kimi is here. workers ai is officially in the big model serving game https://t.co/PbrUcr0jBq - **felixrieseberg** (2026-03-19): https://x.com/felixrieseberg/status/2034688574239776778 - A small ship I love: We made https://t.co/DQ5nmzzS1u and our desktop apps meaningful faster this week. We moved our architecture from SSR to a static @vite_js & @tan_stack router setup that we can serve straight from workers at the edge. Time to first byte is down 65% at p75, prompts show up 50% sooner, navigation is snappier. We're not done (not even close!) but we care and we'll keep chipping away. Aiming to make Claude a little better every day. - **eastdakota** (2026-03-19): https://x.com/eastdakota/status/2034679126951698797 - Prediction: Clay Christensen is best known for his theory of “Disruptive Innovation,” but the business transformations AI is forcing will eventually show that his most impactful work is his theory of “Jobs to be Done.” - **ritakozlov** (2026-03-19): https://x.com/ritakozlov/status/2034673534690271621 - code mode is goat mode. that's the tweet - **threepointone** (2026-03-19): https://x.com/threepointone/status/2034622206882234865 - full rewritten from scratch, now executed (and secured!) inside a dynamic worker, a lot lighter and powerful. much, MUCH more to come. (sharing here early because you know how we do) https://t.co/AAE7IXI65N https://t.co/EiDOODATgf - **zeeg** (2026-03-19): https://x.com/zeeg/status/2034471606014419294 - @thdxr I still don’t know who is and isn’t and boomer and if i am Will AI fix that - **liamihorne** (2026-03-18): https://x.com/liamihorne/status/2034316244816646530 - Cloudflare support for MPP! Add a simple proxy in front of your server to start accepting payments https://t.co/18cnbD6mgV - **ritakozlov** (2026-03-18): https://x.com/ritakozlov/status/2034305672553345474 - great @syntaxfm episode with @southpolesteve on vinext this is one of the best real world examples i’ve seen of ai as an amplifier https://t.co/eXkaQDn5n3 - **stolinski** (2026-03-18): https://x.com/stolinski/status/2034292178806292804 - Cloudflare slop forked Next.js and Vercel lost their minds. Here's how they did it. https://t.co/CuPZQmGFaB - **gregolsent** (2026-03-18): https://x.com/gregolsent/status/2034226168027578795 - I built an LLM judge to score 171k PRs we've merged at intercom over the last 5 years. % of PRs with descriptions communicating the problem and the intent of the changes went down in late 2024 and then took a dramatic nosedive in mid-2025! Guess why? – AI! https://t.co/hiHg0v86iW - **gregolsent** (2026-03-18): https://x.com/gregolsent/status/2034226170737148103 - To fix this I created a Claude Skill that instructs it to figure out the problem statement from the session itself. And, if not clear, ASK the human for the problem statement before creating PR. I was scared rolling this out (Claude asks me a question now?) but folks loved it! https://t.co/RhSFVlHezT - **zeeg** (2026-03-17): https://x.com/zeeg/status/2034044230096851070 - hell yeah https://t.co/YCj0POTbdV - **brian_scanlan** (2026-03-17): https://x.com/brian_scanlan/status/2033978300003987527 - We've been building an internal Claude Code plugin system at Intercom with 13 plugins, 100+ skills, and hooks that turn Claude into a full-stack engineering platform. Lots done, more to do. Here's a thread of some highlights. - **southpolesteve** (2026-03-17): https://x.com/southpolesteve/status/2033977733756117386 - The Pi agent repo closes their issue tracker on weekends. Kinda love this idea. https://t.co/h8wKFNqFtK - **matteocollina** (2026-03-17): https://x.com/matteocollina/status/2033975320869581086 - TanStack Start: 997 req/s throughput, 12.79ms average latency on Watt. 100% success rate. At 1,000 requests per second, average response times stayed under 13ms. This is the performance leader. https://t.co/NAkScfbNDc - **matteocollina** (2026-03-17): https://x.com/matteocollina/status/2033975317384069134 - We benchmarked TanStack Start, React Router, and Next.js running the exact same eCommerce app at 1,000 req/s on AWS EKS. The results were eye-opening. https://t.co/StYgnZZHx1 - **RhysSullivan** (2026-03-17): https://x.com/RhysSullivan/status/2033961101126472151 - my new criteria when evaluating products is if they have a public api that i can give to my agent churning off of my budgeting app because they don't - **threepointone** (2026-03-17): https://x.com/threepointone/status/2033959772882379218 - shoutout to @tomocchino for being a friend and mentor no matter when or where we go - **zebassembly** (2026-03-17): https://x.com/zebassembly/status/2033940530887835856 - Today is my birthday and now I'm halfway through my 20s, wtf - **thdxr** (2026-03-17): https://x.com/thdxr/status/2033920137368101206 - there are so many successful companies in the world they all use different programming languages, tech, processes, hiring strategies, etc and that adds up to whether or not it's fun to work there good thing to remember these days because AI is making people think there's going to be one optimal way of being successful - **dillon_mulroy** (2026-03-17): https://x.com/dillon_mulroy/status/2033744190866149692 - thoughts after day 1 of using pi full time - less is more - i don't miss subagents like i thought i would - /tree is an insanely good context management primitive (and partially why i havent reached for subagents yet) - based only on vibes, i think having a minimum system prompt is improving code quality - telling pi to copy opencodes webfetch and websearch tools was a good play - **thdxr** (2026-03-17): https://x.com/thdxr/status/2033710493114712244 - @zeeg i think 0 to 1 definitely sucks in the way you're describing feature dev in a mature codebase can be good if you aren't lazy which is a huge if - **dkundel** (2026-03-16): https://x.com/dkundel/status/2033638984631615814 - Sorry in a meeting with Codex! ☎️ Voice dictation with Codex in an office is much less awkward this way 😂 https://t.co/IpETeOOGHw - **aninibread** (2026-03-16): https://x.com/aninibread/status/2033626815617765858 - Come find us at booth 735 if you are at NVIDIA GTC this week! https://t.co/1LNDtgKxMs - **michellechen** (2026-03-16): https://x.com/michellechen/status/2033610854994764276 - til jensen pronounces CUDA like could-a vs coo-da - **matteocollina** (2026-03-16): https://x.com/matteocollina/status/2033590130791854167 - Here's the problem. Every time you need to: - Bundle an app into a Single Executable - Run tests without touching disk - Sandbox a tenant's file access - Load AI-generated code at runtime You need the same primitive: a virtual filesystem that hooks into node:fs AND the module resolver. Nobody had built it. - **matteocollina** (2026-03-16): https://x.com/matteocollina/status/2033590128799547597 - @nodejs We're announcing two things: node:vfs, a Virtual File System landing in Node.js core (PR #61478, ~14,000 lines across 66 files) @platformatic/vfs, a userland package you can use today on Node.js 22+ https://t.co/dXD660jJdO - **matteocollina** (2026-03-16): https://x.com/matteocollina/status/2033590124919787819 - .@nodejs has always been about I/O. Streams, buffers, sockets, files. But there's a gap that has bugged me for years: you can't virtualize the filesystem. You can't import a module that only exists in memory. You can't bundle assets into a Single Executable without patching half the standard library. That changes now 👇 - **RhysSullivan** (2026-03-16): https://x.com/RhysSullivan/status/2033365350063198240 - https://t.co/FLuLmCSb5T - **dok2001** (2026-03-16): https://x.com/dok2001/status/2033353095925600401 - . @threepointone is a class act and a pretty cool guy. - **dillon_mulroy** (2026-03-15): https://x.com/dillon_mulroy/status/2033262228674171378 - @dominikkoch lowes is decent https://t.co/yt7eIYk4rO - **irvinebroque** (2026-03-14): https://x.com/irvinebroque/status/2032963459323941193 - @jeff_weinstein has been running since what feels like the dawn of time: https://t.co/RlaNlxeEDu shameless plug: https://t.co/i3QYKX4FeZ - **kimmaicutler** (2026-03-14): https://x.com/kimmaicutler/status/2032925856423882765 - People interviewed in this stories: construction worker, nurse, substitute teacher who commute between East Bay & SF/peninsula. Don’t be cavalier and tank our regional transit system just as gas prices are likely to be elevated for awhile. Vote for @SFBART! https://t.co/EwbTD4RdLN - **morganlinton** (2026-03-14): https://x.com/morganlinton/status/2032836145349013582 - Interesting read on CF workers, something I'm starting to leverage more after learning more about them from Brian. https://t.co/thOErN4Avt - **threepointone** (2026-03-14): https://x.com/threepointone/status/2032836077996695939 - bucket list item ✅: spoke at a jsconf. didn’t suck. https://t.co/zLAFw8Mdnb - **jeremys** (2026-03-14): https://x.com/jeremys/status/2032828719736496183 - A child’s death in Mission Bay is a 5-alarm emergency for the city https://t.co/G4v3ZhGG7X - **akaphill** (2026-03-14): https://x.com/akaphill/status/2032827975071314251 - For those of us using coding agents to write markdown, make slides, or build data analysis as static HTML: sharehtml turns local files into a Google Docs-like experience with real-time comments. Deploy it for your team on your own @CloudflareDev account, protected by Access. https://t.co/HsI3ukIeWF - **elliotarledge** (2026-03-14): https://x.com/elliotarledge/status/2032753593535574433 - https://t.co/SihPQ5dB2K - **danshipper** (2026-03-14): https://x.com/danshipper/status/2032621522280432089 - @mg extremely cool! - **mg** (2026-03-13): https://x.com/mg/status/2032606457544286436 - https://t.co/HHOQqofkm6 - **zebassembly** (2026-03-13): https://x.com/zebassembly/status/2032588322271273237 - Cloudflare is your favorite developer platform's favorite developer platform - **CFchangelog** (2026-03-13): https://x.com/CFchangelog/status/2032586480720265245 - You can now SSH directly into running Containers using Wrangler. Debug processes or run one-off commands without restarting your instance. https://t.co/Xv2beQ9WAU - **RhysSullivan** (2026-03-13): https://x.com/RhysSullivan/status/2032584721876832444 - @threepointone i agree with this so strongly, the amount of functionality that codemode unlocks is insane and untapped right now makes it trivial to do things like workflows, cron jobs, access control scaling AI to orgs, also generative UI is gonna be sick https://t.co/R7CLFsCe8u - **southpolesteve** (2026-03-13): https://x.com/southpolesteve/status/2032584400517443892 - About a year ago I was walking around London with @threepointone talking about how Durable Objects might let us give every human on earth their own little slice of compute. A tiny dedicated personal computer in the cloud. This feels like a glimpse of that future https://t.co/kWPcJ5bo8c - **SophieLWang** (2026-03-13): https://x.com/SophieLWang/status/2032560128906420445 - I made an interactive blog post about how JPEG image compression works: https://t.co/I3HU73R2ae https://t.co/USHl6t9MKG - **dexhorthy** (2026-03-13): https://x.com/dexhorthy/status/2032548724623425675 - lol meta bought a social network for agents like LinkedIn isn’t already leading the category - **kimmaicutler** (2026-03-13): https://x.com/kimmaicutler/status/2032524389758153117 - There's not going to be a revenue base here after they drive ultra HNW to Florida and Texas and keep giving boomers property tax breaks. https://t.co/lUAMUpp3tt - **samlambert** (2026-03-13): https://x.com/samlambert/status/2032518811841343951 - @cramforce RDS proxy is one of our largest sources of leads. it fails constantly. - **southpolesteve** (2026-03-13): https://x.com/southpolesteve/status/2032497569662726553 - To show this off I slopforked the @opencode server so it runs entirely inside a @CloudflareDev Durable Object. No servers. No containers. Hibernates when idle. Try it: opencode attach https://t.co/z0pWVur718 Code: https://t.co/AP1rx5wQ3y https://t.co/Bm5HzP7GQG - **thsottiaux** (2026-03-13): https://x.com/thsottiaux/status/2032326288913084858 - @SIGKITTEN How about we put codex into ChatGPT and then ChatGPT into the Codex that is within ChatGPT - **thdxr** (2026-03-13): https://x.com/thdxr/status/2032317587355226113 - people keep saying "having adhd is an advantage now" no you idiots our attention span is being assaulted non stop and our brains are shriveling and can't focus for shit and instead of confronting this you morons are like "maybe this is good actually" - **thenanyu** (2026-03-13): https://x.com/thenanyu/status/2032270160359260414 - The Codex guys seem like they’re having more fun - **kevinakwok** (2026-03-13): https://x.com/kevinakwok/status/2032252622451982517 - @irvinebroque you are probably over generalizing from being at a very particular company heh - **southpolesteve** (2026-03-12): https://x.com/southpolesteve/status/2032237199769346446 - Some folks asking for vite 8.0 support in vinext. It works already! We've been testing vite 8 in beta since literally day 1. For existing apps "npm install vite@latest". Thats it. Newly initialized vinext apps already use vite 8 once it became latest. https://t.co/fmHBHrLcf1 - **yenkel** (2026-03-12): https://x.com/yenkel/status/2032228590058619324 - @nilslice almost as if a peak of inflated expectations had been reached and we are now disillusioned https://t.co/L3dHL4Y4yd - **ritakozlov** (2026-03-12): https://x.com/ritakozlov/status/2032114033843683646 - at a time when software is abundant, building the right thing is even more critical instead of using it to produce lines of code, you can use it as a really powerful tool for thinking, and sharpening your thinking read for more takes on how AI is changing the PM role https://t.co/HKBMJ7y0aG - **ritakozlov** (2026-03-12): https://x.com/ritakozlov/status/2032109702188838921 - https://t.co/kvYuwFEXb4 - **thsottiaux** (2026-03-12): https://x.com/thsottiaux/status/2031982340826910812 - Hooks are coming to codex. That’s all I wanted to say. - **kentcdodds** (2026-03-12): https://x.com/kentcdodds/status/2031934033219997734 - I love Cloudflare Sandboxes. Saved me a bunch of complexity (and probably money). Check out how my migration went (involves an abandoned approach the agent took that was stupid). https://t.co/ZHyFX3pkMF - **RhysSullivan** (2026-03-12): https://x.com/RhysSullivan/status/2031928761369931981 - MCP sucking is a harness problem, not an MCP problem MCP unlocks behavior that is fundamentally impossible to get via CLI or APIs Bad auth, too much context usage, all get solved with an execution layer - your agent writes code to progressively discover and call tools https://t.co/kUfhhEkiph https://t.co/OEt8e3byYv - **RhysSullivan** (2026-03-11): https://x.com/RhysSullivan/status/2031880575695667261 - the overindexing on CLIs is kind of insane to me it's building a primitive that's not portable, properly discoverable, has no good approval flow DCR / CIMD to APIs would go so much further but CLI is just the current hype thing - **peterschacherl** (2026-03-11): https://x.com/peterschacherl/status/2031767597029396616 - Moved an app to Cloudflare from self-hosting because of performance and the need for high availability. And since I'm using Cloudflare stuff like R2, I can simplify my life with bindings. Also, @PlanetScale via Hyperdrive is insane. - **kimmaicutler** (2026-03-11): https://x.com/kimmaicutler/status/2031732811036848616 - Don’t threaten me with a good time. https://t.co/yEXLmsUxGr - **Cloudflare** (2026-03-11): https://x.com/Cloudflare/status/2031724342803906792 - Cloudflare now returns RFC 9457-compliant structured Markdown and JSON error payloads to AI agents, replacing heavyweight HTML pages with machine-readable instructions. https://t.co/h5toc5NOVt - **brian_scanlan** (2026-03-11): https://x.com/brian_scanlan/status/2031705780261384314 - As of the last two weeks, 90% of pull requests at Intercom are authored by Claude Code. If this is interesting, next month I'm giving a couple of talks at AI Engineer Europe in London, and the DX Summit on this topic. https://t.co/LuMLyAbGiL https://t.co/89g53hprxD - **steren** (2026-03-11): https://x.com/steren/status/2031583629890204020 - Kudos to the folks at @Bloomberg who have been moving JS forward with foundational features. BigInt and Signals come to mind, and now Source Maps https://t.co/VzxyRwW5AK - **thdxr** (2026-03-11): https://x.com/thdxr/status/2031579270389059978 - us: we are struggling to figure out the best way to use coding agents, we don't have clarity yet everyone else: our team is moving at speeds unheard of, all our PRs are ai generated, we've cleared 6 years of backlog man we must really suck huh - **CloudflareDev** (2026-03-10): https://x.com/CloudflareDev/status/2031488099725754821 - Introducing the new /crawl endpoint - one API call and an entire site crawled. No scripts. No browser management. Just the content in HTML, Markdown, or JSON. https://t.co/DCJMek8cZa - **jkup** (2026-03-10): https://x.com/jkup/status/2031432883412279756 - Source maps are a vital part of modern web development, but the lack of a standard made it impossible to add new features or improve the debugging experience. This is the story of how we standardized source maps and how we continue to ship features through standards 📜 🧵 https://t.co/madQtYcDZa - **KentonVarda** (2026-03-10): https://x.com/KentonVarda/status/2031428744003440936 - Man can you imagine how much nicer it would be using GitHub if page loads took <200ms instead of 2-3s? I would be so much happier to use GitHub. - **FredKSchott** (2026-03-10): https://x.com/FredKSchott/status/2031427769775829481 - @astrodotbuild the new cache API /w live content is 🔥 https://t.co/V9ygK3MW88 - **FredKSchott** (2026-03-10): https://x.com/FredKSchott/status/2031420866672865769 - Powered by Flue 🤯 https://t.co/LX3XirSPAh - **astrodotbuild** (2026-03-10): https://x.com/astrodotbuild/status/2031417674119721455 - Astro 6 is here! We completely rebuilt the Astro dev server and build pipeline onto a new, more powerful runtime-agnostic architecture. Plus: New Fonts API, CSP support, an experimental new Rust compiler, and more... https://t.co/ltMsuLM6CD - **_bgiori** (2026-03-10): https://x.com/_bgiori/status/2031412763902534069 - https://t.co/sUV9jpm1wY - **GrantM** (2026-03-10): https://x.com/GrantM/status/2031384778079928712 - In the not too distant future, all software will need to start benchmarking against SOTA models & harnesses for success rate, cost & speed. This will drive the ROI that customers consistently measure against. https://t.co/mC1CbI1uDj - **thdxr** (2026-03-10): https://x.com/thdxr/status/2031377117007454421 - sent this to the team today everything great comes from being able to delay gratification for as long as possible and it feels like we're collectively losing our ability to do that https://t.co/HlIpY86eJn - **TheRealAdamG** (2026-03-10): https://x.com/TheRealAdamG/status/2031217061460893742 - Saint Tibo: Giver of Tokens and Reseter of Limits https://t.co/LRI4sR1y6R https://t.co/JWUn6LNKQv - **bennpeifert** (2026-03-10): https://x.com/bennpeifert/status/2031200425853284578 - very wise friend just sent me this https://t.co/3wIrVWsxgc - **dillon_mulroy** (2026-03-10): https://x.com/dillon_mulroy/status/2031197028865163759 - this deserves more attention, its so good. i've built several worse versions of this over the past few months, mostly as TUIs or nvim plugins, but this nails the ux/dx i want. https://t.co/uWvpZYv36e - **EthanLipnik** (2026-03-10): https://x.com/EthanLipnik/status/2031164433964396627 - Introducing Loom. Loom is a networking framework for Apple devices. It makes it incredibly easy to build apps that have multiple devices communicate with each other. - Peer to peer WiFi direct support - Local network discovery - Unlock a Mac remotely - The same amazing relay system that MirageKit has - iCloud trust + iCloud sharing for friends and family - General purpose - **backnotprop** (2026-03-09): https://x.com/backnotprop/status/2031145299738263567 - Anthropic code review this, clanker review that ... why don't you shut up and review+annotate your own code.... (yes im a loser who still manually reviews code) Originally inspired by a bunch of feature requests and then seeing @dillon_mulroy tweet a similar cool ux. @plannotator for reviewing plans (primary focus) and code, fully oss. OpenCode, @badlogicgames 's https://t.co/tc56yk4Uom, and Claude Code and other clankers - **kentcdodds** (2026-03-09): https://x.com/kentcdodds/status/2031139065589051567 - Finally offloaded ffmpeg from my web server to a separate container. Here's why/what/how: https://t.co/eqAD1VT9bR - **gokulr** (2026-03-09): https://x.com/gokulr/status/2031136588634431937 - Speaking with Head of Product at a large company (30+ PMs). They actively ship code and their expectation is that all PMs on their team do this. Their recommendation for the first 3 things that PMs and designers (and marketing!) should own end to end: 1. Public-facing website 2. Support experience 3. Onboarding / growth flows - **ChromiumDev** (2026-03-09): https://x.com/ChromiumDev/status/2031102566348574984 - Stop slowing down your entire app just to test one API. You can now throttle individual network requests in Chrome DevTools to simulate specific bottlenecks without affecting the rest of your session → https://t.co/IrQ5RDVaTu https://t.co/7KERyf4URr - **kentcdodds** (2026-03-09): https://x.com/kentcdodds/status/2031095250559566062 - This is the first episode generated in a container on Cloudflare using Cloudflare Queues: https://t.co/sSz9LcKZm4 Writing up a blog post about this move soon. - **RhysSullivan** (2026-03-09): https://x.com/RhysSullivan/status/2031087057800700211 - Push your AGI timelines back a week https://t.co/B7xlAzLDjT - **southpolesteve** (2026-03-09): https://x.com/southpolesteve/status/2031086836118859869 - Click here to get access to our code review agent for yourself: https://t.co/GnrV3oxNDb https://t.co/yY9BmmMOUa - **momito** (2026-03-09): https://x.com/momito/status/2031037112795971945 - a bit of an update… excited to share that i’ve joined @Cloudflare to work on the dev docs (and beyond) been watching everyone constantly ship really cool things from the sidelines, looking forward to helping make the docs and devex the best in the world 🌥️ - **doodlestein** (2026-03-09): https://x.com/doodlestein/status/2030968188175806974 - Since tax season is again upon us, I thought I’d share my article from last year about how I use AI to help me do my own taxes now instead of spending over two grand like I used to for error-prone tax prep from a CPA. It mentions Opus 3.7; just use 4.6: https://t.co/RgxBbj7CgH - **romainhuet** (2026-03-09): https://x.com/romainhuet/status/2030817143236042785 - This is winter in San Francisco 🤩 https://t.co/1XMsSAmaVH - **thsottiaux** (2026-03-09): https://x.com/thsottiaux/status/2030798941760147842 - @0thernet Just use Codex. That might have been a single prompt and worked within your $20 sub - **WalshyDev** (2026-03-08): https://x.com/WalshyDev/status/2030773571111301335 - https://t.co/SUWhuYXTLs Blog: https://t.co/cRG5yBdX8J @CloudflareDev Workers Gradual Deployments are seriously underrated and under used, a big reason is there is no managed way to do it. Well, that stops being true today! Add your Worker and start releasing safely - **iamdavidhill** (2026-03-08): https://x.com/iamdavidhill/status/2030642528492441989 - lessons learned building a session feed for an llm: 1. everyone wants a different session feed - **mitsuhiko** (2026-03-07): https://x.com/mitsuhiko/status/2030431346972074489 - NGL, Cloudflare not supporting just hanging out on a subdomain on a Free/Pro plan is really annoying. - **signulll** (2026-03-07): https://x.com/signulll/status/2030404483897815089 - the most underrated hire right now is a great product person. when i say product person i'm def not talking about a product manager. perhaps i think there has to be somewhat of a new role. i don't have a good name for it yet but maybe something like "product thinker".. someone with an intuitive grasp of the product as it exists, where it's soft, where it sings, & how to iterate it toward something even sharper. in some sense, this person has to cohesively hold in their head where this product should be 2 years from now & work backwards from that. i say this cuz when building was hard, engineering was the bottleneck & the status hierarchy often reflected that. building is no longer hard. which means the variance in outcomes has shifted almost entirely to judgment on what to build, how to sequence it, & how to talk about it. & the story matters as much as the thing. internally, it organizes the team around a shared model of why. externally, it shapes the interpretive frame users bring to their first experience. you can't retrofit narrative onto a product & expect it to land, it has to be load bearing from the start. the rarest version of this person sits at the intersection of culture & deep technology. someone genuinely bilingual. they know what's technically possible & they know which cultural currents are real vs. ephemeral. that combo is what separates products that feel inevitable from products that feel assembled. before ppl clap back with this person has always been valuable, i know.. i am just saying now they might be the most *important* person in the room. their value compounds like never before. - **garrytan** (2026-03-07): https://x.com/garrytan/status/2030393709368676554 - A poll of 2,593 Dems confirms the median Dem is moderate and wants effective governance. But an 11% fringe hijacks primaries anyway. San Francisco figured out how to fix it. It starts with one woman who radicalized me. https://t.co/JeGDquIRJQ - **JavierBlas** (2026-03-07): https://x.com/JavierBlas/status/2030376810395861479 - Several Iranian top officials have warned today about water desalination plants (following what they said was an attack to water infrastructure by US-Israel). Re-upping my @Opinion column about why water desalination will be key in the Third Gulf War. https://t.co/XuWT9EYh0U - **karpathy** (2026-03-07): https://x.com/karpathy/status/2030371219518931079 - I packaged up the "autoresearch" project into a new self-contained minimal repo if people would like to play over the weekend. It's basically nanochat LLM training core stripped down to a single-GPU, one file version of ~630 lines of code, then: - the human iterates on the prompt (.md) - the AI agent iterates on the training code (.py) The goal is to engineer your agents to make the fastest research progress indefinitely and without any of your own involvement. In the image, every dot is a complete LLM training run that lasts exactly 5 minutes. The agent works in an autonomous loop on a git feature branch and accumulates git commits to the training script as it finds better settings (of lower validation loss by the end) of the neural network architecture, the optimizer, all the hyperparameters, etc. You can imagine comparing the research progress of different prompts, different agents, etc. https://t.co/YCvOwwjOzF Part code, part sci-fi, and a pinch of psychosis :) - **FredKSchott** (2026-03-07): https://x.com/FredKSchott/status/2030363798679453972 - @threepointone “Libraries don’t make sense any more” is way too extreme lol Great primitives are still essential. If a library gives you a great primitive (Sandbox SDK) that’s A+++. But If a library/framework is encoding business logic into it… idk. Material UI vs tailwind - **threepointone** (2026-03-07): https://x.com/threepointone/status/2030360665324429537 - @DavidKPiano @FredKSchott this doesn't preclude making the libraries a lot thinner/flatter and just presenting much smaller lower level primitives tho? - **RhysSullivan** (2026-03-07): https://x.com/RhysSullivan/status/2030358713564057875 - this is why you should be using Effect, in the long term it's much more token efficient than writing standard TS your AI gets knowledge about how your whole system works, what errors can happen, and can leverage the o11y to debug it all https://t.co/PF4hbrdQTw https://t.co/8wymx4tTpe - **threepointone** (2026-03-07): https://x.com/threepointone/status/2030346844769808598 - .@FredKSchott is convincing me that libraries don't make sense any more, we should publish specs/ implementations, and people implement their own takes given our specs and ideas (which could keep evolving?) it breaks my mental model of deterministic deps, but I'm warming to it... - **dillon_mulroy** (2026-03-07): https://x.com/dillon_mulroy/status/2030118452325654903 - first class protocols look fire https://t.co/xx6VgOYUvr https://t.co/Dqxqq14toE - **signulll** (2026-03-07): https://x.com/signulll/status/2030110932064407805 - if you’re a college student, you can buy: - macbook neo $499 - ipad air $549 - airpods 4 $99 - iphone 17e $599 for ~$1750 you get a full stack of the latest personal hardware that is capable & powerful for all of your needs. that’s absolutely ridiculous. - **chigrl** (2026-03-06): https://x.com/chigrl/status/2030069515728339340 - Delta, American, and United all exited fuel hedging roughly a decade ago, with JetBlue and Frontier also having quit their programs. The rationale for the big three was that their size let them negotiate better fuel prices through volume contracts — hedging premiums weren't worth the cost. In March 2025, Southwest CEO Bob Jordan confirmed the company was discontinuing its fuel hedging program, saying "with the exception of a couple of positive years, it's not been beneficial to the company" - **thsottiaux** (2026-03-06): https://x.com/thsottiaux/status/2030005314737856826 - Codex has been open source from the beginning, and today I am proud that we are able to do more for the community out there. Let's keep building https://t.co/x8NTfJcMr9 - **DarioCpx** (2026-03-06): https://x.com/DarioCpx/status/2029906310330777796 - Let me tell you what’s going to happen this weekend as soon as US stocks trading ends today 1️⃣The US will initiate a large scale air attack - Trump will likely hold a press conference to announce it with big fanfare 2️⃣Iran will be in full defensive posture till sunrise when swarm of drones and likely more advanced rockets will be used in the counterattack: - Attacks on Israel will remain limited and continue to be just retaliation. Iran and Israel are too far from each other geographically and large scale damages don’t carry a significant yield for Iran since Israel strategic military assets are well defended while a big drain on Iran stockpile of drones and rockets - Iran will continue focusing its attacks on US assets located in the ME where the stockpile of air defence rockets is already virtually depleted and its cheap drones can still successfully yield significant damages. - Likely more advanced and powerful rockets will be used to directly target US military assets. - Another couple of tankers belonging to anyone but China and anchored in the vicinity of the Strait of Hormuz will be hit and sunk to remind everyone not to attempt any crossing. 3️⃣ During the weekend another oil infrastructure will be directly hit in any of the ME countries. Iran will continue to deny the attack came from them. 4️⃣ A ground attack will begin on Sunday from Iraq carried out by Iran anti government forces. At the beginning it will find limited resistance and there will be announcements of successful progress. When these troops will be enough inside Iran territory the IRCG counterattack will begin targeting the supply line first and then moving to close the retreat path to envelop enemy forces. US will respond with strong bombing in the area, but aircraft’s will be facing strong air defences. 5️⃣ By the time GLOBEX futures resume trading on Sunday night 6pm EST the intensity of the fights will be limited and the skirmishes will continue for another week with the US keeping a close eye on stocks volatility that must remain as low as possible to avoid a market panic heading into March OPEX next Friday You are welcome 🙏🏻 - **grinich** (2026-03-05): https://x.com/grinich/status/2029648905626079266 - GPT-5.4 just landed, and tonight is Codex Demo Night @WorkOS HQ Come meet fellow builders and see demos of the latest ways that developers are using these models. And DM @georgepickett if you want to demo yourself! https://t.co/KyzqehnK84 - **RhysSullivan** (2026-03-05): https://x.com/RhysSullivan/status/2029630283331555386 - MCP is so back GPT-5.4 has dynamic discovery tool to support thousands of tools - excited to see how this performs in comparison to CLIs which can't be searched https://t.co/AokHVJeXJs - **Osinttechnical** (2026-03-05): https://x.com/Osinttechnical/status/2029606904264007740 - More: https://t.co/yA0OzEa7zG - **thdxr** (2026-03-05): https://x.com/thdxr/status/2029590871646024061 - the biggest tech companies are consumer but a tiny % of people in tech are involved in building consumer products so basically no one gets it and everything you read ends up being wrong - **LongArcNews** (2026-03-05): https://x.com/LongArcNews/status/2029483073897926708 - The Hormuz closure is not a 6-week shock. It is an 18-month stress test for the global food system. Most allocators are trading the obvious: oil up, defense up, gold up. They are missing the connective tissue. 🧵 - **S_OhEigeartaigh** (2026-03-05): https://x.com/S_OhEigeartaigh/status/2029475839654388069 - Real slap in the face to every OpenAI person who spoke up in Anthropic's support. Anthropic staff should be embarrassed - not that this was leaked, but that it was sent. After weeks of calling for industry solidarity, I'm embarrassed. - **grinich** (2026-03-05): https://x.com/grinich/status/2029351354310029619 - 🚨 If you Google “install Claude Code” right now, the top result is malware. Attackers are buying Google Ads for fake Claude Code installers targeting developers. 🤯 https://t.co/HUVYwUhVQX - **RhysSullivan** (2026-03-04): https://x.com/RhysSullivan/status/2029071839880986690 - real MCP has never been tried - **Precigenetics** (2026-03-04): https://x.com/Precigenetics/status/2029037575651119212 - THE most misunderstood part of drug discovery? It is the black box that is the actual cells of the human body. @parmita : 1. Cells are NOT bags of RNA. 2. Cells are DIMENSIONS up from our human brains. 3. The industry is siloed: algorithms vs. data, pharma vs. biotech. https://t.co/ooRKHnLeEX - **dillon_mulroy** (2026-03-04): https://x.com/dillon_mulroy/status/2029007200320897196 - many such cases. it’s never too late to become workflow pilled https://t.co/urfSY9SvU9 - **whoiskatrin** (2026-03-03): https://x.com/whoiskatrin/status/2028909500090986740 - quick ship before bed🎈 you can now stream file changes over SSE with https://t.co/I7Fs4zvrCI() and react in real time https://t.co/GR1Xf7bJ6V - **KentonVarda** (2026-03-03): https://x.com/KentonVarda/status/2028897180149264504 - Sometime in the last couple months AI code review bots got really good. 3-6 months ago they were still posting false positives and sycophancy. Now suddenly I'm getting way better feedback from AI than from humans. - **mattzcarey** (2026-03-03): https://x.com/mattzcarey/status/2028893579628703764 - server side code mode spotted https://t.co/8w6i8oMTfr - **dillon_mulroy** (2026-03-03): https://x.com/dillon_mulroy/status/2028888345116332193 - https://t.co/rc6tlZfJxn https://t.co/DqdNL4HIB3 - **MarcSelwan** (2026-03-03): https://x.com/MarcSelwan/status/2028866670253785375 - The next R2 SQL chagelog is going to be...big. - **thdxr** (2026-03-03): https://x.com/thdxr/status/2028863387979956571 - i love the drizzle team and i'm so happy they're joining planetscale many years ago i came across their team in my replies mentioning their tool they'd been working on it for a while but it didn't have much traction, didn't think much of it one day i did a twitch stream documenting all my issues with prisma and they went over every minute and sent me detailed information on how drizzle solves all of it convinced me to look deeper and i realized they hit the perfect architecture for exactly what was missing from the ecosystem did a call with them and ended up making a video to help get a few more eyes and drizzle was so obviously a good idea that it to blew up immediately and then seeing the team execute so well over the years and continue to be hilarious all while being based out of ukraine is probably one of the most inspiring stories in our space i even remember tearing up reading some of their posts and now it's all come full circle and they're joining planetscale who we've been using for all our projects for almost as long as we've been using drizzle great moment for our industry - **ilyasu** (2026-03-03): https://x.com/ilyasu/status/2028849059570848088 - Why did they make the new Studio Display XDR only 27” 😩 - **ghostwriternr** (2026-03-03): https://x.com/ghostwriternr/status/2028772679726035055 - In the next version of sandbox sdk You can give your agents a full desktop environment like Cursor does, and get them to test their work end-to-end, and show you demos. Made easy for you to integrate with a simple, fully-typed API. https://t.co/tJOwb78N1H https://t.co/5F6RRZaglz - **thdxr** (2026-03-03): https://x.com/thdxr/status/2028661911567409563 - this thing about open source projects making their tests private you're either open source or not - part of that is enabling a successor to show up for net benefit of the world making that intentionally difficult is fine but then it's something different - **thsottiaux** (2026-03-03): https://x.com/thsottiaux/status/2028649088594436225 - Codex is back up. We introduced an issue that caused requests to be blocked and rejected as high cyber risk. This was fixed in ~8 mins and we are now back online and fully operational. Team will reset rate limits, that’s been a while. Enjoy one on us. - **sama** (2026-03-03): https://x.com/sama/status/2028642231138353299 - (I also would like to share this, which I wrote after thinking a little more.) There is a lot we will talk about in the coming days, but since this is one of the first "real deal" decisions we have faced, I wanted to share a few things that have been heavily on my mind the past few days. These are the principles I care most about for this decision: alignment, democratization, empowerment, and individual agency. The democratic process must stay in control, and we must democratize AI. OpenAI should not decide the fate of the world; no private company should. We need to work with governments, but also we need to make sure individuals get increasing power. Things are moving so fast that we need to urgently educate the world so that the democratic process has time to catch up. I think one of our most important strategic decisions ever was the principle of iterative deployment. In particular, the key element required for democracy, such as protection of privacy, must be defended by all of society. I believe that, as some of the creators of this new technology, we deserve to and are obligated to have a loud voice about the risks, pitfalls, and benefits we see. I think we are heading towards a world where the relationship between governments and AI efforts is critical. This will be difficult but it has to happen; I do not see any good future where we don't get there. There should not be games and fights in the press like this; drastic government action should be avoided. I think there are real dangers coming to the world, and maybe pretty soon; I tried to put myself in the mindset of how I'd feel the day after an attack on the US or a new bioweapon we could have helped prevent. - **sama** (2026-03-03): https://x.com/sama/status/2028640354912923739 - Here is re-post of an internal post: We have been working with the DoW to make some additions in our agreement to make our principles very clear. 1. We are going to amend our deal to add this language, in addition to everything else: "• Consistent with applicable laws, including the Fourth Amendment to the United States Constitution, National Security Act of 1947, FISA Act of 1978, the AI system shall not be intentionally used for domestic surveillance of U.S. persons and nationals. • For the avoidance of doubt, the Department understands this limitation to prohibit deliberate tracking, surveillance, or monitoring of U.S. persons or nationals, including through the procurement or use of commercially acquired personal or identifiable information." It’s critical to protect the civil liberties of Americans, and there was so much focus on this, that we wanted to make this point especially clear, including around commercially acquired information. Just like everything we do with iterative deployment, we will continue to learn and refine as we go. I think this is an important change; our team and the DoW team did a great job working on it. 2. The Department also affirmed that our services will not be used by Department of War intelligence agencies (for example, the NSA). Any services to those agencies would require a follow-on modification to our contract. 3. For extreme clarity: we want to work through democratic processes. It should be the government making the key decisions about society. We want to have a voice, and a seat at the table where we can share our expertise, and to fight for principles of liberty. But we are clear on how the system works (because a lot of people have asked, if I received what I believed was an unconstitutional order, of course I would rather go to jail than follow it). But 4. There are many things the technology just isn’t ready for, and many areas we don’t yet understand the tradeoffs required for safety. We will work through these, slowly, with the DoW, with technical safeguards and other methods. 5. One thing I think I did wrong: we shouldn't have rushed to get this out on Friday. The issues are super complex, and demand clear communication. We were genuinely trying to de-escalate things and avoid a much worse outcome, but I think it just looked opportunistic and sloppy. Good learning experience for me as we face higher-stakes decisions in the future. In my conversations over the weekend, I reiterated that Anthropic should not be designated as a SCR, and that we hope the DoW offers them the same terms we’ve agreed to. We will host an All Hands tomorrow morning to answer more questions. - **thdxr** (2026-03-03): https://x.com/thdxr/status/2028621778218557519 - could have prompted a git worktree feature into existence a while ago but we put in the time to design a system that can also support docker containers, remote sandboxes, etc with sync that can tolerate going offline we're always the last to ship, but we're the last to ship https://t.co/T5azGNCXZc - **zorkian** (2026-03-02): https://x.com/zorkian/status/2028611933075906619 - This week, Discord will begin enforcing E2EE on non-stage voice and video calls across desktop, mobile, web, and consoles. Took a few years to solve the edge cases and wait out deprecation times for older clients, but we're here. Really proud of the team today! - **kentcdodds** (2026-03-02): https://x.com/kentcdodds/status/2028590845373391115 - I got some new office equipment. https://t.co/ZLlQd4tuEC https://t.co/4MpBDB7D9G - **StainlessAPI** (2026-03-02): https://x.com/StainlessAPI/status/2028581019343786105 - "SDK code mode" enables operating an API via MCP with SotA accuracy and efficiency. It's how Stainless MCP servers already work today - learn more on the Stainless blog and in the thread 🧵 https://t.co/BI1G4koUtI - **heymingwei** (2026-03-02): https://x.com/heymingwei/status/2028550728370037087 - Trade-secret of the day: add this MCP server and ask anything about Internet traffic trend or routing. It's unbelievably powerful. https://t.co/LZhH6a2MCp https://t.co/KhUvxtv4n3 - **minglu** (2026-03-02): https://x.com/minglu/status/2028546215798087752 - my debut post on the cloudflare changelog. more to come! https://t.co/obxsa97IrN - **michellechen** (2026-03-02): https://x.com/michellechen/status/2028530055438114977 - my contribution to @elithrar's code review bot (bonk) that we use in our @CloudflareDev docs repo. lilbonk uses a semi-secret, highly-anticipated model hosted on workers ai big bonk take lil bonk https://t.co/4Oqf5hsrAq - **grinich** (2026-03-02): https://x.com/grinich/status/2028518179765678427 - https://t.co/KrBfyVdFHZ - **whoiskatrin** (2026-03-02): https://x.com/whoiskatrin/status/2028441553899839755 - Agents SDK v0.7.0 is out 🎈 𝚗𝚙𝚖 𝚒 𝚊𝚐𝚎𝚗𝚝𝚜 • observability rewrite — see exactly what your agent did and when with automatic flow to Tail Workers • keepAlive() — long tasks that actually finish • waitForMcpConnections — no more racing your own tools on startup fixed: ghost messages, silent SQL errors, MCP tools vanishing after hibernation, stream resumption bugs and so much more less debugging. more shipping! - **kylemathews** (2026-03-01): https://x.com/kylemathews/status/2028251456621740214 - I create/invented a new SKILL for leading Claude Code through a structured Hegelian dialectic like process for doing deep investigation of topics. I've spent 40+ hours using it over the past 10 days — super addicting. It's as transformative for me as the original Deep Research. https://t.co/IplZXEmmKe - **threepointone** (2026-03-01): https://x.com/threepointone/status/2028191754910904399 - @PonziChad @cramforce right here! https://t.co/cOGEiHjmlo - **threepointone** (2026-03-01): https://x.com/threepointone/status/2028190927534043630 - this would've been funnier with a Paid partnership disclosure, but I actually mean it! https://t.co/XbjiuoSEug - **threepointone** (2026-03-01): https://x.com/threepointone/status/2028187706786824233 - ok @cramforce you cooked with just-bash. it's very good. - **zebassembly** (2026-03-01): https://x.com/zebassembly/status/2028185702056267838 - One of the best parts of vinext is that when there's a problem that needs to be solved, like commonjs issues, the fix is often using a pre-existing part of the Vite ecosystem. The fix was just adding vite-plugin-commonjs Everything is better with Vite - **natseckatrina** (2026-03-01): https://x.com/natseckatrina/status/2028170853746159961 - @ShakeelHashim It's true. I'm going through the contract now and will push a clear and more comprehensive explanation shortly. - **typesfast** (2026-03-01): https://x.com/typesfast/status/2028166688316031090 - MSC Leeeeeroy Jenkins!! https://t.co/iEcfwoj1Rt - **typesfast** (2026-03-01): https://x.com/typesfast/status/2028133269288665597 - Had a long flight yesterday so I wrote another children's book! Coming soon for Amazon Pre-Order: https://t.co/H5MpnVVDlp - **dok2001** (2026-03-01): https://x.com/dok2001/status/2027934625557815398 - $ 𝚗𝚙𝚖 𝚒 𝚊𝚐𝚎𝚗𝚝𝚜 https://t.co/RnbRiTVgjP https://t.co/CSFOlcbL2j - **natseckatrina** (2026-03-01): https://x.com/natseckatrina/status/2027927506792419626 - A lot of the concerns about the government's "all lawful use" language seem to stem from mistrust that government will follow the laws. At the same time, people believe that Anthropic took an important stand by insisting on contract language around their redlines. We cannot have it both ways. We cannot say that the government cannot be trusted to interpret laws and contracts the right way, but also agree that Anthropic’s policy redlines, in a contract, would have been effective. This is why our approach has been: Let the democratic process decide on the legality and proper use question. The fact that people can even say that the gov has made mistakes in the past is the process in action. The fact that we are having this discussion on twitter is part of the process. Create a reasonable contractual framework that guides expectations and the relationship, just as much if not more than the rules themselves. And on top of this, have the ability to build the models the way we think is safe, along with cleared FDEs to do the real world work in partnership. - **sama** (2026-03-01): https://x.com/sama/status/2027917750858092921 - Enforcing the SCR designation on Anthropic would be very bad for our industry and our country, and obviously their company. We said to the DoW before and after. We said that part of the reason we were willing to do this quickly was in the hopes of de-esclation. I feel competitive with Anthropic for sure, but successfully building safe superintelligence and widely sharing the benefits is way more important that any company competition. I believe they would do something to try to help us in the face of great injustice if we could. We should all care very much about the precedent. I saw in some other tweet that I must not be willing to criticize the DoW (it said something about sucking their dick too hard to be able to say anything critical, but I assume this was the intent). To say it very clearly: I think this is a very bad decision from the DoW and I hope they reverse it. If we take heat for strongly criticizing it, so be it. - **ShrivuShankar** (2026-03-01): https://x.com/ShrivuShankar/status/2027910368316821916 - ok decompiling things with opus is actually very addicting here's apparently how netflix and slack work - https://t.co/oQ5Te9BEtP - https://t.co/MqJy3YzSlO idk maybe wont be that hard to vibe code these after all > use chrome dev tools to explore and provided a grounded tear down analysis of how the site works - **sama** (2026-03-01): https://x.com/sama/status/2027904991953096948 - For a long time, we were planning to non-classified work only. We thought the DoW clearly needed an AI partner, and doing classified work is clearly much more complex. We have said no to previous deals in classified settings that Anthropic took. We started talking with the DoW many months ago about our non-classified work. This week things shifted into high gear on the classified side. We found the DoW to be flexible on what we needed, and we want to support them in their very important mission. The reason for rushing is an attempt to de-escalate the situation. I think the current path things are on is dangerous for Anthropic, healthy competition, and the US. We negotiated to make sure similar terms would be offered to all other AI labs. I know what it's like to feel backed into a corner, and I think it's worth some empathy to the DoW. They are on the a very dedicated group of people with, as I mentioned, an extremely important mission. I cannot imagine doing their work. Our industry tells them "The technology we are building is going to be the high order bit in geopolitical conflict. China is rushing ahead. You are very behind." And then we say "But we won't help you, and we think you are kind of evil." I don't think I'd react great in that situation. I do not believe unelected leaders of private companies should have as much power as our democratically elected government. But I do think we need to help them. - **natseckatrina** (2026-03-01): https://x.com/natseckatrina/status/2027901765698367947 - I'm Katrina Mulligan and I'm OpenAI's Head of National Security Partnerships. I'm excited to be joining @sama for tonight's AMA. - **natseckatrina** (2026-03-01): https://x.com/natseckatrina/status/2027901088582496357 - new account, who dis? - **sama** (2026-03-01): https://x.com/sama/status/2027900042720498089 - I'd like to answer questions about our work with the DoW and our thinking over the past few days. Please AMA. - **thsottiaux** (2026-02-28): https://x.com/thsottiaux/status/2027860345264472549 - Codex is now starting to be associated to speed. Excited for the team to keep building here, there is a lot more we can do and the next few weeks will be fun! https://t.co/QOuTQ5B5HV - **boristane** (2026-02-28): https://x.com/boristane/status/2027858302525444202 - I'm back https://t.co/Zfb73XEbZa - **polynoamial** (2026-02-28): https://x.com/polynoamial/status/2027850059875029335 - For those following the DoW AI drama, I highly recommend reading this post explaining how @OpenAI approached the negotiations with the DoW. - **iamdavidhill** (2026-02-28): https://x.com/iamdavidhill/status/2027848979069739098 - i’ve worked on a lot of products, many of which i never used it is infinitely more enjoyable and valuable for both yourself and your users to work on a product you use daily - **brian_lovin** (2026-02-28): https://x.com/brian_lovin/status/2027839818902991204 - Absolutely insane how much progress you can make on a small side project with a well-made codebase and rigorous verification systems (tests, logging, tracing). - **tbpn** (2026-02-28): https://x.com/tbpn/status/2027775887610179669 - Stripe CEO @patrickc predicts software will shift from "mass-produced, industrial scale" made "years beforehand" to bespoke, custom software created the moment you need it. "Up until now, the economics of software have been conceived of as fixed cost, then infinitely monetized. That has these kinds winner-take-all dynamics." "But once there are inference costs and custom creation involved, it really shifts." - **zebassembly** (2026-02-28): https://x.com/zebassembly/status/2027715235037970555 - about to slop fork the entire ecosystem to be esm - **swyx** (2026-02-28): https://x.com/swyx/status/2027613757787279730 - btw i think a little buried today in the oai fundraise is the fact that OpenAI Codex added 600k users in last 3 weeks: - Feb 4 @sama said it crossed 1M WAU - Feb 27 oai says it crossed 1.6M WAU it is up >3x from Jan 1 (!?!?!?!?!?!?!?) which includes the Codex app launch (Feb 2) https://t.co/Tw3nyFoI4i https://t.co/tv6ArL1908 - **ericclemmons** (2026-02-28): https://x.com/ericclemmons/status/2027601424473858128 - Have y'all seen https://t.co/3TXSqQFZr6?? 👀 Socket+CLI for scripting, too. Let me see what Claude can do with this... https://t.co/vS1oYLxbU1 - **kimmaicutler** (2026-02-28): https://x.com/kimmaicutler/status/2027581902996340865 - PG never misses. https://t.co/B0U00e9wHH https://t.co/rLFM2keXtk - **skeptrune** (2026-02-28): https://x.com/skeptrune/status/2027564083701289160 - you can now use mintlify to automatically build docs for arbitrary github repos (e.g. n8n here) with an agent powered by a skill reflecting everything we know about quality docs simply swap "github" with "mintlify" in the URL for any repo https://t.co/0EW4Jjy2f4 - **alexisohanian** (2026-02-28): https://x.com/alexisohanian/status/2027562624754290895 - The timing of this AI & robotics revolution couldn't be better. In the same way all those developing countries leapfrogged telephone lines by just building cell phone towers, I feel like America is going to be able to leapfrog into robot-led manufacturing 🇺🇸 - **boristane** (2026-02-28): https://x.com/boristane/status/2027555005289664950 - @jshchnz legendary - **cnakazawa** (2026-02-27): https://x.com/cnakazawa/status/2027525188548767799 - Slop this, security that. The underappreciated piece of the story is that Vite is a very good bundler. - **dok2001** (2026-02-27): https://x.com/dok2001/status/2027516276029985130 - How did @thdxr figure out a way to win this week? https://t.co/Qn08SCQhAA - **BraydenWilmoth** (2026-02-27): https://x.com/BraydenWilmoth/status/2027505508890927226 - Slop forks are all the rage these days. https://t.co/mp7lQvnLoF https://t.co/PherjG6KLt - **dexhorthy** (2026-02-27): https://x.com/dexhorthy/status/2027478374520865174 - can we plz settle on what a "harness" is quickly harness == the coding agent (claude code, codex, etc) apparatus == the stuff you build around the coding agent (backpressure, mcp, ralph wiggum, etc etc) i'm tired of this blurry line where ppl use "harness" to mean both the agent and the stuff you build around it - **dok2001** (2026-02-27): https://x.com/dok2001/status/2027449223319351526 - @rauchg @HacktronAI Thay is basically a static site. Stop fear mongering. - **satanacchio** (2026-02-27): https://x.com/satanacchio/status/2027431656513040748 - Yes we need new streams ... again https://t.co/KkyIGRghna - **mjackson** (2026-02-27): https://x.com/mjackson/status/2027430019161854444 - @irvinebroque @kentcdodds @jasnell In particular I strongly agree with the point about using for await … of to consume JS streams. Thats the API that has always felt the most natural to me. We have built in language support for async streams of values. It shouldn’t feel bolted onto the existing spec. - **mjackson** (2026-02-27): https://x.com/mjackson/status/2027422610842353669 - James makes a bunch of great points. I've never been in love with the web streams API. I'm more a proponent of the idea that we should have a streams standard for JS that can be used everywhere. I don't want to write bespoke streaming code that works only in one JS runtime. I was hoping "web streams" would be it, but as @jasnell says all the major JS runtimes have had to work around various issues in the spec so it def has issues. - **ritakozlov** (2026-02-27): https://x.com/ritakozlov/status/2027419886734180858 - starting a movement to reclaim em dashes — they are actually good and not slop! https://t.co/rudWNeDW1Y - **jplhomer** (2026-02-27): https://x.com/jplhomer/status/2027415986328969389 - Ah man, this is so great. Fun to see a breakdown from someone so deep in the weeds of JS implementations. https://t.co/X3MG8JZZNC - **jsneedles** (2026-02-27): https://x.com/jsneedles/status/2027414035046535246 - Sounds a lot like the Pipe semantics in dotnet ❤️ https://t.co/468XXeHpB4 I deal with it a lot 😆 The interesting bit of "There is no partial consumption, no BYOB patterns, no byte-level operations within the streaming machinery itself. Chunks go in, chunks come out, unchanged unless a transform explicitly modifies them." Doesn't really make sense to me - in the web, deserializing the body of the request, you're probably still going to have to deal with partial chunks, since you're not controlling what goes in reallllly? So if you're doing low level json parsing (🙋) - reading in bytes, etc, you still may need to allocate your bytes as u go, because you don't know where a token starts/stops. - **Cloudflare** (2026-02-27): https://x.com/Cloudflare/status/2027404618938884283 - The Web streams API has become ubiquitous in JavaScript runtimes but was designed for a different era. Here's what a modern streaming API could (should?) look like. https://t.co/CgE6E9FW7w - **threepointone** (2026-02-27): https://x.com/threepointone/status/2027399832776659079 - you've got to read this post on a better api for web streams from James Snell, amazing https://t.co/qIeC8VT5Kt - **rough__sea** (2026-02-27): https://x.com/rough__sea/status/2027398866736431220 - Wow this is a huge proposal and analysis from James. I fully agree that WebStreams are an abomination. https://t.co/GdF4hK8fWT https://t.co/FH6n64IwK9 - **dillon_mulroy** (2026-02-27): https://x.com/dillon_mulroy/status/2027361573560353227 - lodash https://t.co/Im6E0MGrL2 - **signulll** (2026-02-27): https://x.com/signulll/status/2027179331554808023 - netflix has to be one of the best run companies on the planet. they’re incredibly disciplined in almost every single way, & likely the most thoughtful in execution. they pumped the price up & forced one of their rivals (paramount) to carry more debt for the acquisition while walking away with a $2.8B breakup fee from the takeover. & the likelihood of netflix acquiring paramount + warner bros in the future as a distressed asset is non zero. stock is up 10%. - **mattzcarey** (2026-02-26): https://x.com/mattzcarey/status/2027119101768831134 - New release of workers-oauth-provider v0.2.4 inc. /.well-known/oauth-protected-resource endpoint, as required by the MCP authorization specification and a bunch more.. npm i @cloudflare/workers-oauth-provider@latest in your MCP servers to upgrade to the latest spec. - **KennyJohnsonATX** (2026-02-26): https://x.com/KennyJohnsonATX/status/2027071843362844762 - ⛳️ SimGolf was one of my favorite childhood games. On my parental leave, I used @claudeai Code, @godotengine, and @Cloudflare to build my own golf course tycoon from scratch in under a month. Open sourced the game and wrote up lessons learned, check it out! https://t.co/GBqavAEEUM - **ritakozlov** (2026-02-26): https://x.com/ritakozlov/status/2027058326002094561 - 👋 i’m the person that put that line in the blog post (i’ve been at cloudflare for 10 years, so i’d say i qualify as old cloudflare). as we say in the post, this came together in a week and we were excited that a real government website (https://t.co/Ulk5PQpNCG) was already running this and things were better for them today on vinext than the day before on opennext intent was not to be misleading or growth hack. it was to show how quickly things can move these days. this IS experimental and it happened to be useful and robust enough for someone to adopt. both things are true albeit a bit nuanced and can see how it comes across - **mitsuhiko** (2026-02-26): https://x.com/mitsuhiko/status/2027052124060742001 - I for one thing that vinext is an interesting experiment, in the same way that just-bash is interesting. Both of those things are implementations that target specs that were created by third party projects. We will probably see more of that, and I think we should. - **joshpuckett** (2026-02-26): https://x.com/joshpuckett/status/2027050667794366718 - @jeff_weinstein Well you already know @brian_lovin - **jeff_weinstein** (2026-02-26): https://x.com/jeff_weinstein/status/2027050145343549902 - who are the best people to follow for agent-first design[0]? [0] design of humans collaborating with agents and building for agents themselves - **levi** (2026-02-26): https://x.com/levi/status/2027042884739125350 - Tried switching from next.js to vinext and my dev reload times went from avg 2-5s to sub 100ms. The triangle company has some explaining to do. - **matteocollina** (2026-02-26): https://x.com/matteocollina/status/2027034178357698864 - If you want to know what pointer compression is about and cut your @nodejs memory in half, you can watch @theo going through what we have built (node-caged) https://t.co/BZOm8e6WPt - **nevikashah** (2026-02-26): https://x.com/nevikashah/status/2027031170320306480 - You can now filter Workers logs and traces with a structured query language. Field queries, regex, boolean logic, and autocomplete built right into the search bar. More investigation 🚀 Less clicks 😮‍💨 h/t @msmps_ https://t.co/omxSK6Fhz5 - **ritakozlov** (2026-02-26): https://x.com/ritakozlov/status/2027015104651485284 - yo we’re just out here shipping… past 2 days at cloudflare: - no config wrangler deploy - sandbox snapshots - pywrangler on windows for python workers - higher resource limits for containers (memory, vcpu, and more instances!) - agents sdk 0.6.0 did i miss anything? 🚀🚀🚀 - **QuinnyPig** (2026-02-26): https://x.com/QuinnyPig/status/2026837110653857996 - I like this. https://t.co/1QqFq98VjQ - **heymingwei** (2026-02-25): https://x.com/heymingwei/status/2026796866181149148 - CPU time reduction after switching to Cloudflare vinext https://t.co/BMB5AbCyb5 - **theo** (2026-02-25): https://x.com/theo/status/2026794317197849001 - Frame․io is falling apart. I wanted something better. Something simple, fast and reliable. I'm excited to introduce lawn․video, an open source video review platform. It flies. This video is not sped up at all. 100 gigs of storage, unlimited seats, $5/month. https://t.co/JOWXxG2mqw - **southpolesteve** (2026-02-25): https://x.com/southpolesteve/status/2026790029901607410 - vinext currently deploys to Cloudflare Workers by default. We want to support all platforms. Proposal for a pluggable adapter system. Feedback welcome: https://t.co/m6McVDDY9G - **dinasaur_404** (2026-02-25): https://x.com/dinasaur_404/status/2026751661838942683 - It's been amazing to support platforms like Mocha —hosting their AI generated apps (100k+), enabling them to do it safely (built in isolation + safely storing secrets) & helping them extend our products (D1, R2) to their end users powered by @Cloudflare Workers for Platforms 🧡 https://t.co/mJdCkmXiWu - **ritakozlov** (2026-02-25): https://x.com/ritakozlov/status/2026739006734766311 - i must confess: something that’s been bothering me for a long time is the experience of deploying an existing project to cloudflare. having to pick framework, configure build, etc this changes today! with autoconfig: just connect to your repo or box wrangler deploy and go! https://t.co/HErh54aPF2 - **dinasaur_404** (2026-02-25): https://x.com/dinasaur_404/status/2026733892204454366 - HUGE — no wrangler file, no problem! when you deploy projects to cloudflare, we'll now automatically detect the framework and configure everything (wrangler file included) that's needed to get your project deployed! https://t.co/fHLaRMXQbO - **yomnashousha** (2026-02-25): https://x.com/yomnashousha/status/2026731045651935680 - Deploying to @Cloudflare Workers just got so much easier 🚀 Bring in any project - even without a wrangler config file - and we'll handle the rest 😏 https://t.co/p1AMbi1K9h - **threepointone** (2026-02-25): https://x.com/threepointone/status/2026674239479169140 - very nice new feature: declare an mcp server, and connect to it with an agent in the same worker/project. purely via rpc, no urls or anything. makes for building personal things really nice. apparently mcp isn't dead after all https://t.co/2CfG7itl9Q - **thdxr** (2026-02-25): https://x.com/thdxr/status/2026673639785910711 - it's so hard to get people to understand that 1000 small swings will never equal 1 big swing - **ritakozlov** (2026-02-25): https://x.com/ritakozlov/status/2026656979704492189 - snapshotting just landed in @cloudflare sandbox sdk - load your environment and dependencies from r2 - reduce cold starts - try it today! and stay tuned for more 👀 https://t.co/nJO9l5OUfY - **whoiskatrin** (2026-02-25): https://x.com/whoiskatrin/status/2026635036078821483 - had a lot of fun shipping this 🎉 cloudflare sandboxes now support createBackup() and restoreBackup(), so you just backup the workspace and instantly restore a ready-to-go environment and this is just the start; native snapshotting at the container level is coming soon 👀 https://t.co/TXO4cuIAER - **miantiao_me** (2026-02-25): https://x.com/miantiao_me/status/2026625162217771111 - 太酷了。 只需一句“migrate this project to vinext”，现在 Agili 的 Hacker Podcast 已经切换到 vinext 了，这 CPU 时间肉眼可见的降低 https://t.co/QoECZrBYTT https://t.co/jeTrqXvctn - **FredKSchott** (2026-02-24): https://x.com/FredKSchott/status/2026438795231703459 - Finally I can re-platform SvelteKit back to Snowpack and return us to the sacred timeline https://t.co/n6UmFrtWBu - **thdxr** (2026-02-24): https://x.com/thdxr/status/2026436386077696015 - pretty crazy this post talks about creating a replacement for OpenNext using OpenCode https://t.co/eKHX88Zvgf - **EricSimons** (2026-02-24): https://x.com/EricSimons/status/2026418811277291661 - WOW. mind blown 🤯 https://t.co/PxZW6i5uM7 - **sebastienlorber** (2026-02-24): https://x.com/sebastienlorber/status/2026417444730925541 - 🤯 https://t.co/QM7N8DOb3f - **wesbos** (2026-02-24): https://x.com/wesbos/status/2026408683991679373 - cloudflare ported Next.js to vite in a week for $1,100 I don't believe it, but excited to try. It's a drop-in replacement, let me know how it works on your app. https://t.co/DZMcE5PTn3 - **jsneedles** (2026-02-24): https://x.com/jsneedles/status/2026405570605994139 - A few months later, hadn't heard a peep from this client. So I went to check on it. All is working wonderfully. Just surpassed 200B events, ingesting ~1.5B per day. Workers as your ingest layer just cannot be beat https://t.co/dir017646T https://t.co/rzOG4MOyNj - **elithrar** (2026-02-24): https://x.com/elithrar/status/2026404330333511978 - $ npx skills add cloudflare/vinext $ opencode run migrate this project to vinext https://t.co/tmpXYRJcFQ - **boristane** (2026-02-24): https://x.com/boristane/status/2026399294870401486 - new front-end framework just dropped $1,100 in tokens we're so back @southpolesteve 🐐 https://t.co/mOAHJfANKn https://t.co/WvqOVoYDkI - **grinich** (2026-02-24): https://x.com/grinich/status/2026398178363482462 - Last week we shipped WorkOS FGA, a more service for defining more advanced and complex permissions. It's one of the most powerful things we've ever built. Check out the full overview here: https://t.co/5dX4RrCdG9 - **evanyou** (2026-02-24): https://x.com/evanyou/status/2026395158762369322 - For the record this is not the thing I've been working on 😂 But very, VERY cool regardless https://t.co/c7oSjLqG3P - **goldmanem** (2026-02-24): https://x.com/goldmanem/status/2026392264462000366 - This is just a peek into the developer platform we’re building at Notion - and we’re starting with agent tools. But soon, everything you can do in the Notion application, you’ll be able to build and extend in code. It’s one small step forward for developers, and a giant leap for coding agents and the promise of truly tailored software 🚀 🌕 - **BraydenWilmoth** (2026-02-24): https://x.com/BraydenWilmoth/status/2026392176301580741 - NextJS cost $1,100 for a developer and AI to rebuild and it's 4.4x faster & 57% smaller. https://t.co/Vvu7Iav5O1 - **goldmanem** (2026-02-24): https://x.com/goldmanem/status/2026392066222354785 - Every Custom Agent deserves a dependable, get-it-done, fully programable, wickedly efficient, CPU partner. So we built what we’re (tentatively) calling Notion Workers. We’re launching Workers in an extremely early alpha today (use at your own risk)! Build your agents the exact tools they need to do work across your business. https://t.co/4f4Klyq715 - **biilmann** (2026-02-24): https://x.com/biilmann/status/2026391766493237642 - This is wild. Looking forward to compare this new vite version of Next on our runtime layer at Netlify as well. The AI rewrite against a huge existing integration test suite is another sign of how coding agents are going to change all the fundamentals of how we build software. https://t.co/emnQvhdUTy - **paulpopus** (2026-02-24): https://x.com/paulpopus/status/2026389779890282813 - So this just drops? I guess let's see how Payload runs on this https://t.co/VfOqTOmzeV - **threepointone** (2026-02-24): https://x.com/threepointone/status/2026389187360944639 - wanna see something cool? agents ⨉ next.js: https://t.co/muauiKTjtO repo: https://t.co/NqU4uLYsPk https://t.co/vRY8zmB5Hl - **caelin_sutch** (2026-02-24): https://x.com/caelin_sutch/status/2026389077113651570 - We deployed this for https://t.co/jPtxqTESOU and has significantly improved core rendering stats: FCP: 396ms (vinext) vs 504ms (live) Eager JS: 85KB / 3 files vs 384KB / 16 files Who could have guessed gov sites could be running cutting edge software https://t.co/MzpJEwALLr - **ndstudio** (2026-02-24): https://x.com/ndstudio/status/2026388957714379173 - Congratulations to @Cloudflare on the launch of vinext! From @Cloudflare: "We've been working with National Design Studio...they're already running vinext in production, with meaningful improvements" - **dillon_mulroy** (2026-02-24): https://x.com/dillon_mulroy/status/2026387490181947878 - you are not, but you can now nextjs powered by vite https://t.co/nc2rr5YQaD https://t.co/0S1eWDOKDS - **ritakozlov** (2026-02-24): https://x.com/ritakozlov/status/2026387150778888252 - we (@southpolesteve) rewrote nextjs on vite in a week and $1,100 worth of tokens.  it’s faster and vendor lock in free, and you can try it with: npm i vinext but my $.02: if your stack is not working for you, in the year 2026, you can probably, actually, just switch to things that do https://t.co/ffTX5rEr8V - **dok2001** (2026-02-24): https://x.com/dok2001/status/2026386974580330830 - It’s Next.js Liberation Day. The #1 request we kept hearing: help us run Next fast and secure, without the lock-in and the costs. So we did it. We kept the amazing DX of @nextjs, without the bespoke tooling, built on @vite. We’re working with other providers to make deployment a first-class experience everywhere. Next.js belongs to everyone. https://t.co/2A3pHnQLfp - **mattzcarey** (2026-02-24): https://x.com/mattzcarey/status/2026380513389236358 - @theCTO 4.6 for vibe, 5.3 for review - **stephenhaney** (2026-02-24): https://x.com/stephenhaney/status/2026348761501581620 - Hello! Today we're releasing Paper Desktop Paper is now a canvas for Cursor, Claude Code, Codex. Any agent can read and write html to Paper. • push or pull from your codebase • pull real data from anywhere • less work, more design What will you ship? Sound on 🎶 https://t.co/2E6OYWpmeP - **andrewchen** (2026-02-24): https://x.com/andrewchen/status/2026338476397076534 - AI is makes software cheaper to build, but not easier to distribute This is the core conflict for years to come - **jhleath** (2026-02-24): https://x.com/jhleath/status/2026285910661194148 - coming soon: archil for mac so you can share the exact same files between agents running in sandboxes and agents running locally https://t.co/TYo5Rwwyr0 - **au6ust** (2026-02-24): https://x.com/au6ust/status/2026178927442686271 - It's been quite a while since I took the time to package up a personal tool. Going to challenge myself to do this monthly for the rest of the year. I present LOOPER, my Ralph loop take that has been yielding pretty fantastic results on long running tasks. https://t.co/rHijQKz2vo - **theCTO** (2026-02-24): https://x.com/theCTO/status/2026107221474685053 - Opus for planning, 5.3 for execution unspoken meta - **soleio** (2026-02-24): https://x.com/soleio/status/2026088755770392902 - Product directors are the convergence of PMs, engineers and designers https://t.co/TEUQarQ9KC - **threepointone** (2026-02-23): https://x.com/threepointone/status/2025998396327399919 - so, "templates" and "starters" are kinda dead yeah? I swear I wanted to write this about 6 months ago but never got around to it. And now, seems foregone. I never liked them, because there was at least one decision that was annoying to undo. But fine, good for a lot of people. And now, even if there's a good starter, I'm going to point my coding boy at it and say "steal the good parts, jut set it up like I always do. Oh and swap out X for Y while you're at it" - **kentcdodds** (2026-02-23): https://x.com/kentcdodds/status/2025992849754652914 - My current biggest bottleneck with AI (other than myself) is agents are only marginally better at CSS than I am (which is bad) and they lack "taste" (whatever that means). They just don't know when a UI looks bad. - **boristane** (2026-02-23): https://x.com/boristane/status/2025984931974033851 - friday was my last day at @cloudflaredev two years ago @dok2001 @ritakozlov @bjyule @irvinebroque @elithrar took a bet on @baselimehq what followed was the most rewarding chapter of my career, shipping with a team that genuinely raised my bar every day grateful doesn't cover it - **_coenen** (2026-02-23): https://x.com/_coenen/status/2025974084664758470 - After waking up to a blizzard this morning, thought it would be a perfect time to re-share https://t.co/xKjEPcaI6G - a wintry mix of the original map. https://t.co/e2HWXOGUfa - **ankrgyl** (2026-02-22): https://x.com/ankrgyl/status/2025639829422448699 - Was talking to someone about "why build your own database" and specifically "how do you beat performance of well known OLAP database X" Here is a good way to think about it -- databases are actually very slow compared to handwritten code. I can write a substantially faster "SELECT COUNT(*) FROM t" with handwritten C++ than the query parsing/optimization/dynamic execution in a database. In fact, some databases (eg Singlestore, Redshift) literally generate and compile code, per plan, to make queries more efficient. So the question is not "how do you perform faster than OLAP database X". It's actually how do you build a system that is sufficiently general for the problem you are trying to solve, without losing too much performance. This is a much harder class of problems. You have to build a really good compiler that inherently supports and optimizes the queries you expect, without over-abstracting (and therefore sacrificing performance) as you add surface area. In the case of braintrust/brainstore (and loosely speaking, observability systems), it's easy for us to beat OLAP databases because they're simply not designed for incredibly sparse and dynamic JSON, full text search on MB+ strings, extremely high cardinality self joins (spans/traces), or running on object storage (at least in OSS). The hard part is adding enough surface area to support the features we want to expose in the UI and giving our users general enough SQL. - **willccbb** (2026-02-22): https://x.com/willccbb/status/2025406019669766214 - fake software engineering jobs are dying rapidly. real software engineering jobs, however, - **EastlondonDev** (2026-02-21): https://x.com/EastlondonDev/status/2025325347747848390 - @mattzcarey @whoiskatrin I have no idea what is going on here 😂 Is @irvinebroque going to Lisbon any time soon, he can bring a hat - **jarredsumner** (2026-02-21): https://x.com/jarredsumner/status/2025306979812642974 - feature request for GitHub’s gh cli: a subcommand for agents monitoring PRs - view unresolved pr review comments as markdown or xml with file:line - show failing gh action logs, maybe filtered to near the error - show lint errors from gh actions - **thdxr** (2026-02-21): https://x.com/thdxr/status/2025284706586026312 - @dharmesh wow that looks amazing did you guys build that? - **thdxr** (2026-02-21): https://x.com/thdxr/status/2025280623775043957 - whenever people talk about "shipping speed" they focus on 0 to 1 but what actually matters is how fast the team is still shipping a year into the project this is why everyone seems like they "ship fast" but nothing actually seems to be getting done - **zeeg** (2026-02-21): https://x.com/zeeg/status/2025273169930842122 - when did silicon valley become a bunch of cosplay founders or was it always this way and I just chose to ignore it? - **boristane** (2026-02-21): https://x.com/boristane/status/2025261675197424001 - the software development lifecyle is dead requirements → design → code → test → review → deploy → monitor this loop is finished I wrote down my thoughts https://t.co/T56Eq4z3ef - **steipete** (2026-02-21): https://x.com/steipete/status/2025196111598666177 - @shl @daviduche03 People act as if using agents to build software removes all the work; it accelerates the process, but it also upped expectations. - **karimatiyeh** (2026-02-21): https://x.com/karimatiyeh/status/2025181417500541141 - Silicon Valley was so ahead of its time 😂 https://t.co/luuvUtqpxw https://t.co/SfNLuFKI7M - **ritakozlov** (2026-02-20): https://x.com/ritakozlov/status/2024945460478111785 - love this detail on the workers product page: https://t.co/5F7kEDobOp perfect illustration of why cloudflare workers is so well suited for the age of endless execution environments (think: infinite agents, code mode, etc) https://t.co/COvfrnY5X6 - **mitsuhiko** (2026-02-20): https://x.com/mitsuhiko/status/2024895741370458549 - This is how MCP should work! https://t.co/xMwYnwSpR6 - **threepointone** (2026-02-20): https://x.com/threepointone/status/2024888612878163993 - We (and by we I mean @mattzcarey ) shipped something else today that’s potentially a bigger deal and I’m just waiting for the mcp buzz to settle in before going all in on talking about it Weekend blog incoming https://t.co/YOpvbOLw1I - **beyang** (2026-02-20): https://x.com/beyang/status/2024885811154743639 - This is the way. With current model capabilities, MCP is less about defining tools and more about defining the auth and code execution substrate. There will be more things that fit this pattern beyond MCP. You had kernel and user space—now we have agentspace. https://t.co/PtHlDkBobI - **michellechen** (2026-02-20): https://x.com/michellechen/status/2024871394530734488 - i don't believe in 1m context window and i was right https://t.co/2erK3XYHa0 - **caelin_sutch** (2026-02-20): https://x.com/caelin_sutch/status/2024871297809834088 - Very cool - congrats to the CF team excited to use this https://t.co/WsdDApItxH - **StainlessAPI** (2026-02-20): https://x.com/StainlessAPI/status/2024866731722768565 - MCP is approaching an inflection point, and the limiting factor is becoming obvious. Last week we held our first event in SF with @KeycardLabs, @Cloudflare, @temporalio, and @humanlayer_dev. We talked about what’s breaking in MCP right now, and what it will take to make agent-driven software robust. Some hard truths: - Many MCP servers work in demos but fail across real-world use cases. - “Giant agent skills” are becoming an anti-pattern with thousands of lines explaining things the model already knows, without clarity on the 5% it doesn’t. - Standard MCP approaches (one tool per endpoint or dynamic tools) scored ~50% correctness, but more advanced implementations reached ~70%. - The real kicker: when agents wrote code against production-grade SDKs, correctness jumped to 90–95%. But capability is only half the story. Control is the other half. An in-house Gmail MCP server was great for drafting emails… until it sent one unintentionally. The question isn’t just “can the agent do this?” but “what is it allowed to do?” OAuth scopes and coarse endpoint controls aren’t enough, and we’ll need fine-grained, programmable boundaries like: - Limit an agent to a single document - Cap refund amounts - Prevent broad, unintended actions If AI interfaces are going to replace dashboards, the underlying systems need to behave like an operating system kernel, not a shell script. Capability gets attention. Control earns trust. Trust is what scales. - **dinasaur_404** (2026-02-20): https://x.com/dinasaur_404/status/2024856336064724999 - this is crazy cool 🤯🤯🤯 cloudflare's entire API surface (1000+ APIs) as an MCP server powered by code mode 🧡 freaking crushed it @mattzcarey https://t.co/7NAhcKFawJ - **thomas_ankcorn** (2026-02-20): https://x.com/thomas_ankcorn/status/2024853810326192496 - One MCP to rule them all, One MCP to find them, One MCP bring them all and in the Worker bind them https://t.co/cJSdLPOnC9 - **kentcdodds** (2026-02-20): https://x.com/kentcdodds/status/2024853273060999224 - This guy is a wizard and must be followed https://t.co/2bN6J53wqf - **kentcdodds** (2026-02-20): https://x.com/kentcdodds/status/2024852855610380454 - Moving Code Mode from a client concern to a server concern is brilliant. Very excited about this idea for MCP servers with large surface areas to cover! https://t.co/o2LPXm06WE - **elithrar** (2026-02-20): https://x.com/elithrar/status/2024851906305118243 - MCP is dead (long live MCP) https://t.co/h65rX3dS4i - **dillon_mulroy** (2026-02-20): https://x.com/dillon_mulroy/status/2024851567334011110 - i told you mcp is so back https://t.co/UfB6F0fyqJ - **threepointone** (2026-02-20): https://x.com/threepointone/status/2024849445611872667 - Start the discourse. We’re claiming to have fixed the context problem with mcp. (Well, a lot of it). Lots to say about this, but this marks a new branch of the tech tree opening up. Code mode for everything. You heard it here first. https://t.co/XxVJsBFu7Z - **Cloudflare** (2026-02-20): https://x.com/Cloudflare/status/2024847784914882945 - The Cloudflare API has over 2,500 endpoints. Exposing each one as an MCP tool would consume over 2 million tokens. With Code Mode, we collapsed all of it into two tools and roughly 1,000 tokens of context. https://t.co/rpWBqGao0a - **ritakozlov** (2026-02-20): https://x.com/ritakozlov/status/2024847632217063831 - we just launched the new @cloudflare mcp server! it has only two tools: search and execute and reduces the number of input tokens used for context by 99.9%, compared to equivalent native MCP implementation powered by server-side code mode; powered by dynamic worker loader! https://t.co/tKh6egJ5IL - **mattzcarey** (2026-02-20): https://x.com/mattzcarey/status/2024847630811980277 - Code Mode is all you need, very excited about this direction for MCP https://t.co/a5E5Xqrvrt https://t.co/gkmcRrRl1I - **ChatGPT** (2026-02-19): https://x.com/ChatGPT/status/2024630205973991782 - https://t.co/kXHrvTEDBJ - **pitdesi** (2026-02-19): https://x.com/pitdesi/status/2024609108566167588 - IMO the BEST sandwiches and dips in SF are at a cute little bodega near Dolores Park: Dolores Deluxe Founder Ramzi has a great story- he trained at Eleven Madison Park, and got there via WeightWatchers, his first job. After a promotion at WW, and his boss took him to EMP to celebrate. He was so blown away by the meal that he quit WW to go work there, initially checking coats, before working his way up and moving through other Guidara restaurants. My favorite sando is the Lady Giana: caramelized eggplant, basil, mozzarella, fontina charred tomato arrabbiata + garlic potato chips (inside!) on acme focaccia garlic bread 🤤 They don’t do delivery platforms at the moment so you have to go in person. Try it, say hello to Ramzi, and grab a cookie! - **kentcdodds** (2026-02-19): https://x.com/kentcdodds/status/2024582755674444119 - I've been building things using cloudflare's workers platform and it's been great. https://t.co/SOEmh3ioa8 - **MikeNomitch** (2026-02-19): https://x.com/MikeNomitch/status/2024550435059879963 - Easiest way to get an Agent that... - Replies to chat, email, cron, API calls, more - Has tool calling, workflows, Human in the loop, browser control - Stores its own state in 1 LOC - Integrates with MCPs, or has its own MCP server - Can run and host the code it outputs https://t.co/p27Twiiwm3 - **threepointone** (2026-02-19): https://x.com/threepointone/status/2024546932626952622 - https://t.co/Xn4CJJAHJS (beta) a demo of every agents sdk feature. it's a lot. we're lacking features from other frameworks/infra, and we'll close that gap very quickly. but there are features in our agents sdk that will never be in other frameworks/infra. https://t.co/9pMRHyDPnE - **boristane** (2026-02-19): https://x.com/boristane/status/2024525111928508837 - @thorstenball okay, but what are you building? - **jlongster** (2026-02-19): https://x.com/jlongster/status/2024524508380398032 - very cool idea opencode does that I haven't really seen before: beta version is auto built from latest dev branch AND includes all PRs with the `beta` label so you don't need to merge anything to get some real testing for it, you just apply the `beta` label - **LakeAustinBlvd** (2026-02-19): https://x.com/LakeAustinBlvd/status/2024496479801421910 - biased: come join @Cloudflare Lisbon, our loveliest office unbiased: come join @Cloudflare Workflows, a key piece of the most exciting work underway and do it sitting next to the phenomenal @efalcao and team here https://t.co/XGGMBdvHJo - **boristane** (2026-02-19): https://x.com/boristane/status/2024320768189092021 - notion but it's just markdown - **brian_lovin** (2026-02-19): https://x.com/brian_lovin/status/2024312228628111448 - @benisahuman @Cloudflare Awesome work, are we going to publish more blog posts about these types of improvements? I'd love to learn - **akaphill** (2026-02-19): https://x.com/akaphill/status/2024295110998200807 - I wanted native iOS notifications when my background agents (and workflows) finish or need approval. Tried Twilio and my account got suspended right after signing up? Ended up building my own. Paje is a simple iOS app + self-hosted API on Cloudflare Workers. Agents make an HTTP POST and you get a notification on your phone. - **benisahuman** (2026-02-19): https://x.com/benisahuman/status/2024290205168259416 - Notion just got way faster! We now stream HTML from the edge → we're seeing the speed of a static site while keeping dynamic content A @Cloudflare Worker starts streaming the <head> while fetching preload data, so by the time your browser boots Notion, the data's already there https://t.co/sDcoX2UrBE - **thomas_ankcorn** (2026-02-19): https://x.com/thomas_ankcorn/status/2024273028314542331 - Cloudflares clickhouse infra will never stop blowing my mind. 1.61 quadrillion events per day…. The team that work on this are so legendary https://t.co/Y39i3GBPFl - **southpolesteve** (2026-02-18): https://x.com/southpolesteve/status/2024204861466624088 - If OpenCode adds markdown file viewing to the sidebar I will live stream myself uninstalling obsidian https://t.co/wzSbfKiyE9 - **dok2001** (2026-02-18): https://x.com/dok2001/status/2024193685030916470 - . @lalkaka and I spent years as part of the product team on-call at Cloudflare, this is exactly what we wished existed. Congrats on the launch — the world needs fewer dashboards and more software that just works. https://t.co/VLl2A17MgL - **elithrar** (2026-02-18): https://x.com/elithrar/status/2024175550168752307 - very excited to see the Firetiger team launch! also great see a built-in adapter for Workers observability 😎 https://t.co/EPu2O15jRI https://t.co/2wpNCxiA8i - **kamathematic** (2026-02-18): https://x.com/kamathematic/status/2024172377689514413 - few things more soul crushing than setting up github actions - **lalkaka** (2026-02-18): https://x.com/lalkaka/status/2024169210545672556 - Today we're introducing @usefiretiger. You and your AI agents write code. Firetiger makes sure it works. Our team and I have plenty of incident war stories building @Cloudflare, @segment, @Twitch. In the agentic coding era, the volume of code changes + quality issues in prod is ever increasing, but observability vendors aren't incentivized to close the gap. They make money when you write more data to them, not when your software actually works. Firetiger is the agentic operations layer for the agentic coding era. We combine production observability data, codebase understanding, and knowledge of your business to find problems before your customers do and fix them before they notice. We've raised $7.6 million led by @sequoia with participation from angels who believe in better software, including @eastdakota, @calvinfo, @NicoRosberg, @dok2001, @jeffawilke, and @alanaagoyal. You can sign up for @usefiretiger today, self serve. We charge for agents that directly make your software better and more reliable, not for observability data ingested, with plans starting at $599/month. Observability is dead. Long live outcome engineering. - **elithrar** (2026-02-18): https://x.com/elithrar/status/2024108033341620352 - we’ve been using our OpenCode-based bot for docs & code review over the last few weeks but found ourselves needing to repeat certain things. so I told it to look at the PRs where @irvinebroque + I had steered it + build all of it into the new skill: https://t.co/h02a0iDg10 - **morganlinton** (2026-02-18): https://x.com/morganlinton/status/2023957763010023849 - I'm starting to get into a better groove with GPT-5.3-Codex for overnight builds. This is the key sentence that's working for me right now: "define a strict build contract (deliverables, tests, reports, and stop conditions) so it’s clearly an overnight-sized feature before coding" I then review this, make any tweaks and adjustments, have it run again, and don't stop until I feel like it actually has something that will actually take the whole night to code. - **draecomino** (2026-02-18): https://x.com/draecomino/status/2023939529057726906 - I've followed tech for 25 years and I've never felt a larger gap between the ~1 million people using Codex/Claude and the rest of humanity. - **thomasgauvin** (2026-02-18): https://x.com/thomasgauvin/status/2023918307720503579 - Developer meetup for the @CloudflareDev data platform: talking R2 SQL, Apache Iceberg and Lakehouses with @cloudera and @lancedb! https://t.co/BpPLtLvNVn - **matthewcp** (2026-02-17): https://x.com/matthewcp/status/2023904371591766425 - Constraints are with us every step in API design, yet most people are not even aware of it. If you're willing to face the constraints you place on yourself, you free yourself to see the full gamut of possibilities. https://t.co/IJ7czyY2s6 - **ritakozlov** (2026-02-17): https://x.com/ritakozlov/status/2023879509548667242 - this is so cool — @laravelphp cloud now supports markdown for agents! feb 13 customer request -> feb 17 launch 🚀 powered by @cloudflaredev https://t.co/IlDQqCh47C https://t.co/CX33ykqCjl - **zebassembly** (2026-02-17): https://x.com/zebassembly/status/2023877255051825249 - @thomas_ankcorn @WalshyDev I'm pretty sure our bots are just quantized walshy distillations - **laravel** (2026-02-17): https://x.com/laravel/status/2023873477040902626 - Laravel Cloud 🤝 Agents Markdown consumes 80% fewer tokens than HTML. Make your site agent friendly with a single toggle in Cloud. https://t.co/XEutbcU4gE - **thomas_ankcorn** (2026-02-17): https://x.com/thomas_ankcorn/status/2023867915813220500 - Without @WalshyDev Cloudflare would be measurably worse in so many ways https://t.co/vGzz0K7BOx - **alexalbert__** (2026-02-17): https://x.com/alexalbert__/status/2023834863858769975 - Underrated dev upgrade from today's launch: Claude's web search and fetch tools now write and execute code to filter results before they reach the context window. When enabled, Sonnet 4.6 saw 13% higher accuracy on BrowseComp while using 32% fewer input tokens. - **ritakozlov** (2026-02-17): https://x.com/ritakozlov/status/2023830939004993607 - the @cloudflare plugin for @cursor_ai just launched! this builds on top of cloudflare skills, making it really easy to ship applications with the @cloudflaredev ecosystem. you really really can just ship things now! https://t.co/hgCLES58cM - **cursor_ai** (2026-02-17): https://x.com/cursor_ai/status/2023827897627377783 - Integrate with Cloudflare’s developer platform, build MCP servers, and manage Workers with the Cloudflare plugin. https://t.co/Poac39Cuab - **steipete** (2026-02-17): https://x.com/steipete/status/2023818964602687734 - In der USA sind die meisten Menschen enthusiastisch. In Europa werde ich beschimpft, Leute schreien REGULIERUNG und VERANTWORTUNG. Und wenn ich wirklich hier eine Firma baue dann kann ich mich mit Themen wie Investitionsschutzgesetz, Mitarbeiterbeteiligung und lähmenden Arbeitsregulierungen abkämpfen. Bei OAI arbeiten die meisten Leute 6-7 Tage die Woche und werden depentsprechend bezahlt. Be uns ist das illegal. - **ankrgyl** (2026-02-17): https://x.com/ankrgyl/status/2023810273598128588 - We sent this note to our customers to let them know that Braintrust has raised a new round of funding, and thank them for their support. While the money is exciting, our focus hasn't changed: we're building Braintrust to help our customers ship quality AI products. In 2026, AI is moving to production but teams have never had less conviction about what will fail next. Our customers are building AI products that serve millions and simply need to work. If Braintrust makes their lives easier and their products better, I know we are doing our job. Thank you to @ICONIQCapital for leading our Series B, and to @a16z, @GreylockVC, @basecasevc, and @eladgil for doubling down. Thank you to the Braintrust team for all the incredible work you've done over the past year. And thank you to our customers, who have made this growth possible. - **matteocollina** (2026-02-17): https://x.com/matteocollina/status/2023805916961800301 - @nodejs V8 has had pointer compression since 2020 (Chrome uses it!), but Node.js couldn't enable it — because all worker threads shared a single 4GB memory cage. @Cloudflare sponsored @Igalia to fix it in V8 with a new feature called IsolateGroups. - **jasonlk** (2026-02-17): https://x.com/jasonlk/status/2023802014090883165 - https://t.co/BGw2mMXn0c - **levie** (2026-02-17): https://x.com/levie/status/2023635013963559352 - In a world of openclaw, codex, claude code/cowork, manus, and other agentic systems, it’s becoming clear that the future of software has to be API-first, but also enable human interaction for verification, collaboration with agents and people, and working on the output. It’s generally been the case that software was built for people first and foremost, and then APIs are exposed for other systems to connect into that tool or data. But if we imagine a world where AI agents are doing 10X or 100X more work with software than people, then this paradigm is flipped. Software becomes API-first, with ways of having humans be able to work effectively with the agent, either through a UI as relevant, or chat. If you’re not API-first, then you’re nearly DOA to agents. - **dillon_mulroy** (2026-02-17): https://x.com/dillon_mulroy/status/2023566483784134947 - https://t.co/BBlVj9RAfX https://t.co/5ipfhAbNfk - **celso** (2026-02-17): https://x.com/celso/status/2023559383729488154 - Markdown for Agents keeps getting better: - Larger HTML documents. - Compressed origin responses now supported. - Higher compatibility. More coming. https://t.co/nUhaisG3hg - **wesbos** (2026-02-16): https://x.com/wesbos/status/2023541811491512520 - Been using cloudflare tunnels to expose local servers to the internet, gated with access (GitHub login) Today i setup the opposite, vpn access into my home network with zero trust. Certainly use cases for both, but I’m curious what your default would be for a single app? https://t.co/WGYlbRKs6a - **tobi** (2026-02-16): https://x.com/tobi/status/2023517423761322157 - Running a company is just context engineering internally. Now that skill has even more value in the agentic world. Us tech founders have been doing reps to prepare for this. - **southpolesteve** (2026-02-16): https://x.com/southpolesteve/status/2023516567271862672 - February 16, 2026 - **dbeyer123** (2026-02-16): https://x.com/dbeyer123/status/2023514692959060325 - https://t.co/aDQrH4aqUH - **tobi** (2026-02-16): https://x.com/tobi/status/2023511827116945617 - @elkelk this is false. source: running $150b company - **ctatedev** (2026-02-16): https://x.com/ctatedev/status/2023504884914344348 - The SDLC is evolving bc of agents We've hit a point where well written Issues are often more valuable than PRs PRs are always appreciated, esp for complex work But a clear, reproducible bug report or a focused feature proposal with defined goals is usually faster to ship It keeps review light and aligned on safety, quality and scope - **dexhorthy** (2026-02-16): https://x.com/dexhorthy/status/2023474794017288322 - the best ai engineers I know focus on backpressure and verification https://t.co/wz0NY3kxuc - **tobi** (2026-02-16): https://x.com/tobi/status/2023466837623881848 - Being "creative, but easily distractible" is no longer a disadvantage with infinite agents on your screen - **cryps1s** (2026-02-16): https://x.com/cryps1s/status/2023441322838028362 - Introducing Lockdown Mode for ChatGPT. Lockdown mode is an advanced, optional security setting for higher-risk users, businesses, and enterprises. Lockdown Mode disables certain tools and capabilities in ChatGPT that an adversary could attempt to exploit to exfiltrate sensitive data from users’ conversations or connected apps via attacks such as prompt injections. This includes moving to cached web browsing, and preventing interactions with the web more broadly. While not designed for the average user, I think this will really help protect a lot of organizations who demand strong, deterministic security controls for their workflows. The team really did a great job with this one. Available for Enterprises/Businesses, with consumer access to follow soon. https://t.co/6QViNRmPGS - **lateinteraction** (2026-02-16): https://x.com/lateinteraction/status/2023437567786692966 - @dexhorthy oh that's a whole direction of probing internal representations through tokens that ~no one is thinking about i wonder whether GEPA-like algorithms are the right tool here; maybe training a small prompter with RL can work better because it's not a very verbalizable search space - **dexhorthy** (2026-02-16): https://x.com/dexhorthy/status/2023433160470773774 - nerd snipe for my optimization friends - can you use GEPA to create a secret language that Claude models understand but GPT models don’t? - **can** (2026-02-16): https://x.com/can/status/2023423761513685135 - it’s crazy that a thunderbolt kvm switch is more expensive than a Mac mini - **KentonVarda** (2026-02-16): https://x.com/KentonVarda/status/2023423366150119761 - I submitted a bug report for a minor UI annoyance in @opencode on Saturday night, thinking "no one will care". @nexxeln had a PR up in 3 hours, @adamdotdev merged shortly thereafter, and it landed in a release Sunday around noon. What. 🤯 https://t.co/6Gp9jzRAz4 - **eastdakota** (2026-02-16): https://x.com/eastdakota/status/2023417691860849100 - @nytimes This is incorrect. Cloudflare had no outage. Check your facts before you publish incorrect information. - **sorenblank** (2026-02-16): https://x.com/sorenblank/status/2023214633298039149 - prediction cone/safe triangle — this is something we take so much for granted in modern day native UIs. but it's not the same for most web-based dropdown menus. it took me a while to implement this here. Amazon, macOS, Windows all implement some version of this. https://t.co/jG9pqLWxCE - **davemorin** (2026-02-15): https://x.com/davemorin/status/2023171737949442301 - 🦞 Been working with Peter Steinberger (@steipete) on the OpenClaw Foundation structure for weeks. A home for thinkers and hackers and those that want to own their data. Honored to serve as the founding independent board member. This community built something extraordinary, our job is to protect it. Open source forever. Excited to share more soon. - **iannuttall** (2026-02-15): https://x.com/iannuttall/status/2023162842547998855 - Jan 27th: rebrand from Clawdbot to Moltbot (now @OpenClaw) after Anthropic send a legal letter Feb 15th: 19 DAYS LATER ACQUIHIRED BY OPENAI... A generational fumble by Anthropic here. https://t.co/Aeco8eW9Tg - **raphaelschaad** (2026-02-15): https://x.com/raphaelschaad/status/2023158707329400966 - I bet this causes lots of VC tears and angry OSS folks. But think about this: - Peter showed the future and lots of awesome startups are starting to bloom from this. Invest in those! - Peter created one of the most exciting OSS projects in years. The community is vibrant and alive, it can’t disappear! Also hilarious outcome given the naming journey. Congrats to OAI and @steipete on this move. - **steipete** (2026-02-15): https://x.com/steipete/status/2023154018714100102 - I'm joining @OpenAI to bring agents to everyone. @OpenClaw is becoming a foundation: open, independent, and just getting started.🦞 https://t.co/XOc7X4jOxq - **ctatedev** (2026-02-15): https://x.com/ctatedev/status/2023142972578308546 - Weekend project: portless Stable, named URLs for local dev → Zero config → Subdomains for monorepos → No more port roulette or cookie bleed → Agents get a reliable URL instead of guessing ports 𝚗𝚙𝚖 𝚒 -𝚐 𝚙𝚘𝚛𝚝𝚕𝚎𝚜𝚜 https://t.co/WzkA7xRMCK - **ljharb** (2026-02-15): https://x.com/ljharb/status/2023072415249793047 - @saltyAom a promise doesn’t “run”. A promise is a placeholder for the future value of work that’s already begun. The thing that began the work is what would need to be cancelled. - **steipete** (2026-02-15): https://x.com/steipete/status/2023062172860141642 - @iwantlambo yah they only sent love letters from legal - **steipete** (2026-02-15): https://x.com/steipete/status/2023057089346580828 - PRs on OpenClaw are growing at an *impossible* rate. Worked all day yesterday and got like 600 commits in. It was 2700; now it's over 3100. I need AI that scans every PR and Issue and de-dupes. It should also detect which PR is the based based on various signals (so really also a deep review is needed) Ideally it should also have a vision document to mark/reject PRs that stray too far. This can't be fully automated, but even assisting would help. The closes I found is an obscure oss project. How's no startup working on this? - **eastdakota** (2026-02-15): https://x.com/eastdakota/status/2023042135939825949 - I got to have dinner with Massimo Bottura at his incredible restaurant in Modena a couple years ago. He said something that’s stuck with me: there are only two styles of cuisine in the world, Chinese/French & Japanese/Italian. The first is all about technique and layers. The second is all about quality ingredients and simplicity. - **_chenglou** (2026-02-15): https://x.com/_chenglou/status/2023033064163815472 - The real question I have here is if language's just a really, really long-lasting and valuable premature optimization. The visual cortex bandwidth is so much wider but we throw away so much of it under the premise that we should compress "noise" away. Are they noise just because we treat them as brittle details? Can a better brain capture the full visual scene of a split second & unlock much better reasoning? QR code is "noise" to us but is both robust and high-info for machines - **hiaaryan** (2026-02-15): https://x.com/hiaaryan/status/2022985705862139984 - a few days ago, i promised something. introducing → Sileo ✨ an opinionated physics based toast library, with SVG filters and performance optimisations. try it here: https://t.co/pET5CQKxpx happy sunday ✌️ https://t.co/V4AJx9cpkl - **threepointone** (2026-02-15): https://x.com/threepointone/status/2022959885823144095 - I ran this on the agents package and... wow. I work with the smartest people. https://t.co/zHoNLjbBDW https://t.co/AVlbqgu1IB - **garrytan** (2026-02-15): https://x.com/garrytan/status/2022944652375265382 - AI just moves the excuses out of the way. You don’t get to say “I can’t edit” “I don’t know how to code” or “I’m not technical” Intelligence is officially too cheap to meter. Your taste is your evals. Your desire to win is your agency. Time to build. https://t.co/kXaYlP9TQU - **thdxr** (2026-02-15): https://x.com/thdxr/status/2022916571744866683 - what if in opencode.json you could specify project references references: ["git@github.com:Effect-TS/effect.git"] these would get cloned to a global cache and kept updated and opencode would have a subagent that could answer questions about them would you use this? - **sarthakgh** (2026-02-15): https://x.com/sarthakgh/status/2022908565208346850 - “Cognitive debt is likely a much bigger threat than technical debt….” https://t.co/6eCn4hB5rO “As generative and agentic AI reshape how software is built, understanding and managing cognitive debt may be one of the most important challenges our field faces.” https://t.co/WIdIkyGQFY - **thdxr** (2026-02-15): https://x.com/thdxr/status/2022901590496018708 - everyone tries to predict the future and position themselves to benefit meanwhile someone has to actually make that future happen - **jarredsumner** (2026-02-15): https://x.com/jarredsumner/status/2022898432424505830 - @RhysSullivan I think PR-scoped code coverage is also a good idea but haven’t merged this in to bun test yet Basically: if the new changes do not get run in the tests it wrote, CI should fail - **docmilanfar** (2026-02-15): https://x.com/docmilanfar/status/2022888464556134482 - this is so cool and terrifying: +80mph-80mph = 0mph https://t.co/FZtks9pCZX - **ryancarson** (2026-02-15): https://x.com/ryancarson/status/2022888381823242594 - Getting the entire Harness Engineering system setup for my repo and it's blowing my mind. It even identifies when UI was changed and records a video of testing in the browser and adds those videos to the PR for me to review. I'm using this setup as a guide: https://t.co/u5f9wF4Ssv - **dexhorthy** (2026-02-15): https://x.com/dexhorthy/status/2022877616160530648 - yes we’ve been calling these “golden context” (facts about the system) and “golden rules” (coding/arch guidelines) And while docs can get out of date, if you keep this high level can concise enough you can inject them into every context window and not have to worry they’ll get out of date - **eshear** (2026-02-15): https://x.com/eshear/status/2022859161743344127 - One of the biggest future industries: education. Energy, compute, and data can now be converted into arbitrary skill against a known rubric with ever-increasing efficiency. Education is how you convert skill into taste. - **matthewcp** (2026-02-15): https://x.com/matthewcp/status/2022857347454955632 - When We Stop Reading Code https://t.co/eAbO0svXFk - **vkhosla** (2026-02-15): https://x.com/vkhosla/status/2022855731934814513 - yes https://t.co/X6S3e46W92 - **BorisMPower** (2026-02-15): https://x.com/BorisMPower/status/2022830932449751537 - LLMs enumerate the space of thought, and RL optimizes it - that’s kinda just it ! https://t.co/2Big7yS7F1 - **yenkel** (2026-02-14): https://x.com/yenkel/status/2022819475289002314 - heard from a founder with a strong team working on low level systems: “guess who the top bug finder on our team is? claude” most haven’t caught on yet - **RayDalio** (2026-02-14): https://x.com/RayDalio/status/2022788750388998543 - https://t.co/tjmbT5ytUN - **elithrar** (2026-02-14): https://x.com/elithrar/status/2022786729162260599 - @irvinebroque @dillon_mulroy try now: https://t.co/cfs0jn8JqQ https://t.co/1k4mCpbdsq - **0xblacklight** (2026-02-14): https://x.com/0xblacklight/status/2022750710308442468 - this is 100% true btw but at the same time, all the best engineers I know on this train are doing something so far from removed from what "big vibe coding" is pushing https://t.co/s1dbMNjtl5 - **ankrgyl** (2026-02-14): https://x.com/ankrgyl/status/2022739819273806135 - programming does not really feel dead to me i just spend a lot more time thinking about tests. and evals :) - **southpolesteve** (2026-02-14): https://x.com/southpolesteve/status/2022734569901035605 - Claude Code added SSH but it doesn't work with OSX targets. Codex hasn't added SSH at all yet. I bet if this gets enough retweets I can get these teams to compete on the feature over the weekend. - **lateinteraction** (2026-02-14): https://x.com/lateinteraction/status/2022725370152190215 - RLMs are not sub-agents or the ability to iteratively retrieve context. I know because I trained multi-hop models for reasoning & retrieval in 2020, including compaction.* RLMs are the simplest/purest scaffold that understands its own prompts via recursion, not via attention. They support an extremely simple but unusual claim: Models need to be able to access their own conversations with the user and their own horizon symbolically and recursively. The model should be only allowed to understand this long context by *writing code* that launches LLMs, and composing these into the final response. Note that the number of LLM launches can be linear or even bigger in the context size, not a small constant number of sub-tasks. This sounds big until you remember that attention is already quadratic. I'll have to confess that I always found (and still find) the conventional pattern of "sub-agents" rather boring. This is the superficially related structure where the model is given a special tool it can invoke by writing out prompts for and receiving the output. Verbalizing specific individual sub-calls as tool calls token-by-token hides the internal reasoning from the main context, which is an OK outcome for sub-task delegation. But it's a completely unrelated pattern to teaching models to understand their own context/horizon recursively. Sorry I'm a bit of a pedant for understanding concepts precisely, but this seemed needed. *The title is quite literally "Robust Multi-Hop Reasoning at Scale via Condensed Retrieval", arXived on Jan 2nd 2021. It could work for many steps, retrieve text from a massive corpus, compact/condense its own context, and iterate further. - **antranapp** (2026-02-14): https://x.com/antranapp/status/2022716048227905938 - I can't recommend App Store Connect CLI by @rudrank enough. I'm using it more and more in my workflow now. Just migrated my workflow to upload a build using to testflight from fastlane to ASC CLI 🤟 Github: https://t.co/7Urd8ks6Me https://t.co/skLlZc3BSQ - **iamharaldur** (2026-02-14): https://x.com/iamharaldur/status/2022704544833028133 - **swyxio** (2026-02-14): https://x.com/swyxio/status/2022592190174900463 - **swyx** (2026-02-14): https://x.com/swyx/status/2022580899737673810 - https://t.co/RbcWvwbWwa - **thdxr** (2026-02-14): https://x.com/thdxr/status/2022562561732812841 - v1.2.0 of opencode includes our migration to sqlite - if anything goes wrong don't worry, your original data is not yet deleted you can clear out `opencode.db*` files in %APPDATA% or ~/.local/share/opencode folder and it'll rerun this paves the way to a lot of cool features! - **swyx** (2026-02-14): https://x.com/swyx/status/2022529414547017757 - Am judging @hackwithtrees Stanford sama’s takes on: - the literal first day of OpenAI - “if you’re a sophomore, you’ll graduate to a world with AGI” - why @MattPRD’s Moltbook matters - what you should work on inspiring in how down to earth it sounds but still extremely impactful when done well. - **_chenglou** (2026-02-14): https://x.com/_chenglou/status/2022489656168976405 - Stupidly late realization on why LLMs are so good at reasoning: human’s reasoning capability is bottlenecked by language! It’s not that languages are good at reasoning; reasoning ended up being defined by language first and foremost. The medium truly shapes the message https://t.co/oNcNKobtCx - **zebassembly** (2026-02-14): https://x.com/zebassembly/status/2022479354706694524 - time to vibe code swift workers https://t.co/9MCbfGSG6x - **swyx** (2026-02-13): https://x.com/swyx/status/2022449122469646556 - re: app layer > model layer one structural advantage of why agent labs/open source hackers like @openclaw can run rings around the infinite money glitched model labs of the world is 1) they can simply argmax(all model labs) 2) they can yolo permissions without safetyniks being able to block anything as vocal a fan i am of claude cowork, i think this is a nice small area where non-biglab people can outwork and outexplore capabilities. - **jeff_weinstein** (2026-02-13): https://x.com/jeff_weinstein/status/2022446501671354858 - 📬 Okay, spam is totally out of control. I want to charge $0.0X for people who aren't in my contacts to appear in my inbox. (At least I'd get paid for seeing stuff!) We now have the machine payments @stripe infra to power it. @gmail (cc @sundarpichai) want to build it together? - **ankrgyl** (2026-02-13): https://x.com/ankrgyl/status/2022440201457668226 - pre-vibing, as a human, you do more work then the consumers of your work (eg code reviewers, users, etc) post-vibing, the consumer does more work than you we need to figure out better vibing workflows to restore the balance - **matt_slotnick** (2026-02-13): https://x.com/matt_slotnick/status/2022428696595108152 - https://t.co/LoAntSjTRR - **jhleath** (2026-02-13): https://x.com/jhleath/status/2022401213631713566 - last fun AWS story. in 2016, we'd just launched EFS and someone came to us asking for a massive 15 GiB/s of throughput (lol). we told them to stop using the service. who was that customer? openai. or, as we knew them in 2016, "elon musk's AI thing" i wonder how much butterfly effect that single decision had for the industry - **ericzakariasson** (2026-02-13): https://x.com/ericzakariasson/status/2022392681805348914 - https://t.co/lrbAJfctnR - **mitsuhiko** (2026-02-13): https://x.com/mitsuhiko/status/2022377436042633438 - Today @thorstenball said he's the bottleneck. I too am the bottleneck and it made me think. https://t.co/FIwf1zAMU7 - **CloudflareDev** (2026-02-13): https://x.com/CloudflareDev/status/2022365942899187737 - New: @zai_org GLM-4.7-Flash on Workers AI, @cloudflare/tanstack-ai for @tan_stack AI, and workers-ai-provider v3.1.1 with transcription, TTS, reranking & streaming fixes. Build AI agents entirely on Cloudflare 🔥 https://t.co/6XYoeiiUfb - **deloreyjmi** (2026-02-13): https://x.com/deloreyjmi/status/2022362000265818293 - We've got intern slots in Lisbon on the Workflows team. Apply and send me your Workflows projects! https://t.co/mXlpUBdEGN - **threepointone** (2026-02-13): https://x.com/threepointone/status/2022358006588149882 - This release is special to me. The tweet buries the lede, read the linked post. - A good model on workers ai. - a fully featured adapter for tanstack/ai for all of workers ai (you know I love @tan_stack ) - a completely rewritten workers-ai-provider (for @aisdk, also a fan), zero breaking changes - both libs now get brand new hooks for our transcription/tts/embedding models (in addition to chat, images) We spent the past couple of weeks fixing tons of bugs in our libs related to streaming, tool calling, rendering, just everything. It’s gone from being janky to very very smooth. Every model shines now. If you’d tried workers ai via the ai sdk provider, you should try again. Here, try this playground; https://t.co/2c1pzaURBR Tanstack fans, here’s a playground for you: https://t.co/gVNU6NNyA2 Same consideration of quality. It’s so good. Just like the tanstack. Zero api keys from inside a worker. One key, if you’re using it from somewhere else. All the models. This is part of a bigger plan we’re knee deep in executing right now. - **ritakozlov** (2026-02-13): https://x.com/ritakozlov/status/2022355645677109684 - shipped on just a friday at @cloudflaredev: → GLM-4.7-flash on workers ai (great for tool calling!) → @cloudflare/tanstack-ai: adapter for ai gateway and workers ai → new capabilities in workers-ai-provider: transcription, speech to text, & reranking (and many other improvements!) → (free!) origin CA support for workers vpc always be shipping - **Youkhna** (2026-02-13): https://x.com/Youkhna/status/2022350920516882878 - I built a Mac app that tells me exactly what to focus on the moment I open my laptop. No notifications. No chaos. Just clarity. This is what your computer should feel like. https://t.co/XIaRn1Gun4 - **dok2001** (2026-02-13): https://x.com/dok2001/status/2022339274767520246 - Everything we're doing to make codebases "agent-ready" (better docs, less dead code, smaller surfaces) engineers always needed too. Agents just have zero tolerance for the entropy humans learned to work around. They can't "just know" a file is outdated or a code path is dead. They take your codebase at face value, which means it finally has to be worth taking at face value. - **thomasgauvin** (2026-02-13): https://x.com/thomasgauvin/status/2022337670568108425 - already have a server configured with cloudflare tunnel + origin certs and want to access it via Workers VPC? this one's for you 👇 https://t.co/lm22oVhYVJ - **thdxr** (2026-02-13): https://x.com/thdxr/status/2022313736518688981 - it's crazy how good our brains are at noticing patterns we all instantly pick up on "that's not x — it's y" as llm speak without ever consciously registering this pattern or being taught it - **thdxr** (2026-02-13): https://x.com/thdxr/status/2022308281255919976 - i've always been pro making money, being money motivated and if i tried hard to do something great then money would be a side effect it feels like silicon valley lately is infected by a different version of this where the money is the entire point very uninspiring - **LakeAustinBlvd** (2026-02-13): https://x.com/LakeAustinBlvd/status/2022247631896936710 - Google’s AI tools are 30 different teams in a trench coat. It’s always: > Introducing Thing by Google! Does it integrate w/ Antigravity? > Of course not. Isn’t this Opal? > We don’t know, that’s another team. Can this edit Drive files? > Better than you could ever dream. But consumer only, no workspace support. - **sqs** (2026-02-13): https://x.com/sqs/status/2022242892237164716 - If you haven't changed at least 50% of your direction/roadmap/plans in the last 4 weeks, you are making a very risky choice. - **dok2001** (2026-02-13): https://x.com/dok2001/status/2022156234917589034 - Intern-ception @Cloudflare today: an ex-intern turned founder negotiating with an ex-intern turned PM to use a new product. Couldn’t be prouder of @DhravyaShah @michellechen Not too late to apply for this summer! https://t.co/cEXbcw3CAN - **thdxr** (2026-02-13): https://x.com/thdxr/status/2022118213266628964 - i have been selling software for a long time the amount of enterprise inbound we're seeing and the massive sums of money they're ready to throw at us is unbelievable feels good to know opencode will be self sustainable even with a big well paid team - **boristane** (2026-02-13): https://x.com/boristane/status/2022116258033152080 - missing the east coast rn https://t.co/eBz6iyWsHb - **latentspacepod** (2026-02-13): https://x.com/latentspacepod/status/2022105955039608871 - From rewriting Google’s search stack in the early 2000s to reviving sparse trillion-parameter models and co-designing TPUs with frontier ML research, Jeff Dean has quietly shaped nearly every layer of the modern AI stack. As Chief AI Scientist at Google and a driving force behind Gemini, Jeff has lived through multiple scaling revolutions from CPUs and sharded indices to multimodal models that reason across text, video, and code. We sat down with Jeff to unpack what it really means to “own the Pareto frontier,” why distillation is the quiet force behind every generation of faster, cheaper models, how energy not FLOPs is becoming the true constraint on AI compute, what it takes to co-design hardware and models 2–6 years into the future, why unified multimodal systems will outperform specialized ones, what it was like leading the charge to unify all of Google’s AI teams, and his prediction that deeply personalized models with access to your full digital context will redefine what useful AI looks like. @JeffDean @GoogleDeepMind @Google - **boristane** (2026-02-12): https://x.com/boristane/status/2022038209329565990 - https://t.co/4ezI1MQ3gI https://t.co/8X0IZLChSS - **kathyyliao** (2026-02-12): https://x.com/kathyyliao/status/2022028819532853740 - ~80% reduction in token usage for agents just by adding this header -H "Accept: text/markdown" And for accessing websites not on Cloudflare, you can always use the Browser Rendering /markdown REST API to render a dynamic page or application in a real browser before converting https://t.co/rpIY3ValSt - **boristane** (2026-02-12): https://x.com/boristane/status/2022017414469496943 - one THOUSAND tokens per second ? https://t.co/AimolGDLJc - **sama** (2026-02-12): https://x.com/sama/status/2022011797524582726 - GPT-5.3-Codex-Spark is launching today as a research preview for Pro. More than 1000 tokens per second! There are limitations at launch; we will rapidly improve. https://t.co/Havtaxficn - **OpenAIDevs** (2026-02-12): https://x.com/OpenAIDevs/status/2022009906329739681 - Introducing GPT-5.3-Codex-Spark, our ultra-fast model purpose built for real-time coding. We’re rolling it out as a research preview for ChatGPT Pro users in the Codex app, Codex CLI, and IDE extension. https://t.co/6knTmyQZ4N - **zeeg** (2026-02-12): https://x.com/zeeg/status/2022005042581278857 - been up late routinely and its finally kicking my immune systems ass both love and hate for how much fun building has become again - **dinasaur_404** (2026-02-12): https://x.com/dinasaur_404/status/2021999222393032790 - automatically supported for ✨platforms✨ so you can enable it for all your customers or select custom hostnames https://t.co/caq7YGPJUr - **burcs** (2026-02-12): https://x.com/burcs/status/2021997042487427274 - last week we opened kumo to the public it's been a labor of love and a ton of fun to build. lots of behind-the-scenes work going on to make it more powerful and easier to use you can now theme and match your brand just by updating base tokens can't wait to see what you build! https://t.co/vtat4t6Y8l - **kristianfreeman** (2026-02-12): https://x.com/kristianfreeman/status/2021973423027892625 - In beta now: 1. Turn Markdown for Agents on in CF dashboard 2. Agents make requests with `Accept: text/markdown` to your site 3. CF auto-converts your content to Markdown before returning it to your agents This is awesome. Read more: https://t.co/rlSQrtTTHo https://t.co/6vxZzB3H4I - **boristane** (2026-02-12): https://x.com/boristane/status/2021970366525280431 - markdown is the language of llms and now you can convert your website to markdown with a single click in cloudflare https://t.co/Ge1rscSYrs - **ritakozlov** (2026-02-12): https://x.com/ritakozlov/status/2021968338675478991 - markdown is the language of agents and becoming the new language of the web! we made it one click to make sure your website is speaking it ✨ https://t.co/wyql5dSref https://t.co/p4g754NHAg - **Cloudflare** (2026-02-12): https://x.com/Cloudflare/status/2021955521213800489 - Time to consider not just human visitors, but to treat agents as first-class citizens. Cloudflare’s network now supports real-time content conversion to Markdown at the source using content negotiation headers. https://t.co/B7wYH4PtA8 - **CloudflareRadar** (2026-02-12): https://x.com/CloudflareRadar/status/2021948140169973964 - What kinds of content do #AI bots & crawlers consume? New Content type data on Radar shows the distribution. Apply bot, crawl purpose, industry, and vertical filters in the Data Explorer. Individual bot pages show the distribution for that crawler. https://t.co/eF94aoip3s - **celso** (2026-02-12): https://x.com/celso/status/2021913483227681049 - Big https://t.co/Sqdp4Im10w - **jarredsumner** (2026-02-12): https://x.com/jarredsumner/status/2021902459170697381 - I don’t love the UX of worktrees co-located LLM transcripts feel like the feature of the version control system that replaces git. commits & tags are zoom levels for context. I think PRs and CI as this step completely independent from local dev doesn’t make sense anymore - **RhysSullivan** (2026-02-12): https://x.com/RhysSullivan/status/2021779023580611066 - playing around with showing running agents in the notch https://t.co/mxHPFLN4tc - **nembal** (2026-02-12): https://x.com/nembal/status/2021778181829931422 - talking about .agent identity ;) @ianlivingstone @RattrayAlex @irvinebroque @dexhorthy @andrewtorkbaker https://t.co/PPXuW3STZ9 - **jeff_weinstein** (2026-02-12): https://x.com/jeff_weinstein/status/2021774744031572046 - after a few weeks away, even in the rush hour traffic, in the rain, under the glow of incomprehensible ai ads, i love the bay. - **martin_casado** (2026-02-12): https://x.com/martin_casado/status/2021757905461293339 - @JustJake Extremely well handled. You and team do a world class job in a very very hard space. - **JustJake** (2026-02-12): https://x.com/JustJake/status/2021749053529411784 - Today we failed some of our users Due to a change in fraud rules, ~3% of workloads were incorrectly paused We take this deeply seriously, and extend our sincerest apologies to those affected Full post mortem below https://t.co/bmk6NZzBfY https://t.co/mHUYfYPRku - **thdxr** (2026-02-11): https://x.com/thdxr/status/2021709299010220342 - lol people replying "YOU'RE WRONG OPUS IS MORE ENJOYABLE" uhh yeah that's the point here - **boristane** (2026-02-11): https://x.com/boristane/status/2021692467805082088 - every limit on the cloudflare developer platform getting lifted https://t.co/h2SU4eFNWN - **thdxr** (2026-02-11): https://x.com/thdxr/status/2021674924360831353 - codex is by far a better coding model than opus - anyone who knows anything understands this but the whole industry should reflect on why opus is the most popular people assume whatever is the smartest will win but the old rules of product are still what determine everything - **MikeNomitch** (2026-02-11): https://x.com/MikeNomitch/status/2021657686585360413 - Thrilled about this change to Workers. Each Workers request can now make WAY more subrequests. Previously, it was capped at 1000, which could be tough if you had long-running Websockets or Workflows. Not an issue any more... https://t.co/OkUXaANrrs - **willccbb** (2026-02-11): https://x.com/willccbb/status/2021653634828972447 - with the latest models, i am now finding myself thinking about complex system design problems more, not less. the magnitude of what can be reasonably attempted is monumentally larger. you need to make sure you’re asking it to build the right thing. - **zeeg** (2026-02-11): https://x.com/zeeg/status/2021653390980522025 - If you build iOS apps and you haven’t tried out XcodeBuildMCP you’re sorely missing out. Cameron has excellent taste and built a truly useful tool that makes some of the least pleasant parts of developing in Xcode - particularly with agents - way less tedious. https://t.co/QInwl6Y4I8 - **boristane** (2026-02-11): https://x.com/boristane/status/2021628652136673282 - plan mode sucks, across all coding agents I wrote down my workflow for working with claude code: - plan in a dedicated doc - annotate the doc - iteratively work with claude with a persistent artifact that doesn't get compacted https://t.co/0Zz68i1M68 - **yuris** (2026-02-11): https://x.com/yuris/status/2021617972562006138 - COBOL maintainers: https://t.co/iyQSRl6rdu https://t.co/hZ5YcH01Gm - **dinasaur_404** (2026-02-11): https://x.com/dinasaur_404/status/2021582967642951910 - today's my 7 years at cloudflare and honestly every day still feels like its own adventure. i'm beyond grateful for this place, the people who work here, and all the crazy problems we get to solve. here are 7 of my favorite projects that i've gotten to work on 👇 - **kepano** (2026-02-11): https://x.com/kepano/status/2021563951532974483 - It's not because your agent can do something with 12 tool calls and 20k tokens that it should. For example, Obsdian CLI has a command called "orphans" that finds files with no incoming links. Now your agent can get a deterministic answer in 1 millisecond instead of 4 minutes. - **martin_casado** (2026-02-11): https://x.com/martin_casado/status/2021386139555475831 - I’m so impressed with 5.3 Codex. Well done to the @OpenAI team. I’d run into diminishing returns with 4.6 primarily on backend state management tracking which was becoming whack-a-mole of breaking shit when updating the schema and front end handlers. But 5.3 has been fantastic for sane architecture and separation of concerns. - **zebassembly** (2026-02-11): https://x.com/zebassembly/status/2021379477457039674 - my favorite Cloudflare behavior is shipping so fast that you build something cool before I can even learn about it, many such examples https://t.co/yJtSEBJgiM - **boristane** (2026-02-10): https://x.com/boristane/status/2021369029894832535 - ui looks familiar https://t.co/J0XryZiwnj - **swyx** (2026-02-10): https://x.com/swyx/status/2021362455168024649 - single most important career advice of my last 40 years seems to consistently be "always reflect where the smart/ambitious people are and just dive in" physical/mental nexus is king - **miacmal** (2026-02-10): https://x.com/miacmal/status/2021346873584517560 - logs are now available directly in the Queues dashboard 📃 @CloudflareDev https://t.co/DdMzI59dBy - **__morse** (2026-02-10): https://x.com/__morse/status/2021332462312489439 - https://t.co/N9tfTGfVcb - **__morse** (2026-02-10): https://x.com/__morse/status/2021332459934318918 - I am working on a TUI for gmail, powered to opentui https://t.co/Im8xjbPpws - **jeff_weinstein** (2026-02-10): https://x.com/jeff_weinstein/status/2021331763960873058 - Autonomous agents are an entirely new category of users to build for, and, increasingly, to sell to. Today, we’re launching (a preview) of machine payments on @stripe—a way for developers to directly charge agents, with a few lines of code. 🤖💸 $ Let’s start tinkering… ⤵️ - **Baconbrix** (2026-02-10): https://x.com/Baconbrix/status/2021286101378650574 - In the next version of bacons/apple-targets, Over 20 new apple extensions — totaling 45 ways to extend your app across the Apple ecosystem. ◆ Custom keyboards ◆ Apple Watch complications ◆ Safari/SMS adblock https://t.co/hgrfKW3NT7 - **Claudia_Sahm** (2026-02-10): https://x.com/Claudia_Sahm/status/2021273952706810187 - WH keeps 'pumping up' payrolls for tomorrow. Here's why. Revisions could wipe out all job growth last year. Zero. zip. nada. https://t.co/RILDizthdl - **KeycardAI** (2026-02-10): https://x.com/KeycardAI/status/2021269129861193884 - We acquired https://t.co/HDCWaMzCYx. Keycard now powers autonomous coding agent workflows. Developers shouldn't have to choose between agent capability and security. 🧵 https://t.co/d2FarhXNoO - **garrytan** (2026-02-10): https://x.com/garrytan/status/2021260964838940789 - A developer used OpenAI’s codex to port the entire SimCity (1989) C codebase to TypeScript in 4 days—without reading a single line of code. This is vibe coding at scale. The brave new world awaits your prompt https://t.co/NJ3RmSpmAi - **grinich** (2026-02-10): https://x.com/grinich/status/2021255841194627317 - I wrote the book I always wanted. 📖 A complete, practical guide to enterprise readiness — how to go from “users love us” to “enterprises buy from us.” 275 pages. 18 chapters. The missing manual for founders going upmarket. Limited print run. Want a copy? https://t.co/lzD5uiFNEQ - **chris_mccord** (2026-02-10): https://x.com/chris_mccord/status/2021254850332205465 - non elixir minds cannot comprehend just rpc'ing prod and introspecting state/debug inside the running vm. Now attach that to an AI agent (w/ human in the loop) and you have insane capability. "xyz seems bugged, use fly ssh -C ... to poke the release to debug it" No MCP required. - **dillon_mulroy** (2026-02-10): https://x.com/dillon_mulroy/status/2021243739633570272 - entropy https://t.co/BBP09y2j9E - **kentcdodds** (2026-02-10): https://x.com/kentcdodds/status/2021222535585726672 - MCP things keep happening exactly as I predicted them to do and it's making my bets I made last year look smarter all the time. In this case it's that AI agents would evolve into browsers and browsers would evolve into agents and computer use would work but be a stopgap until a more efficient MCP-based mechanism became available. - **nevikashah** (2026-02-10): https://x.com/nevikashah/status/2021216308012978310 - 🙈🤭 this is just one of MANY new workers observability features in the works. Can you guess the others? https://t.co/iovv60GEis - **betomoedano** (2026-02-10): https://x.com/betomoedano/status/2021192426061193258 - Expo Router just unlocked a new level of native on iOS 🚀 For years, cross-platform meant tradeoffs. Your app worked everywhere, but it never felt quite right. That gap is finally closing. With the latest version of Expo Router, native iOS APIs are now exposed declaratively. No UIKit wrestling. No fake UI. If you want your app to feel truly native on iOS, these are the APIs you should be using: - Native Tabs (real system tab bar) - Stack Toolbar (native header controls) - Bottom toolbars with placement=“bottom” - Bottom accessories attached to the tab bar - Zoom transitions for fluid navigation - Native sheet footers (experimental) I wrote a full breakdown with code examples and video demos of these features in my blog (link below 👇) - **theo** (2026-02-10): https://x.com/theo/status/2021124850191155377 - Last week: "wow everyone who had 5.3 early is obsessed, they must have been paid off to act like this" This week: "...oh" - **theo** (2026-02-10): https://x.com/theo/status/2021122658092580914 - So many of my core beliefs about software development are crumbling in front of me in real time - **dillon_mulroy** (2026-02-10): https://x.com/dillon_mulroy/status/2021105127856406703 - lmao opus 4.6 has started writing python scripts on its own to correctly align ascii diagrams https://t.co/xOJjIQLfup - **US_Stormwatch** (2026-02-10): https://x.com/US_Stormwatch/status/2021082422939677051 - California’s Sierra Nevada could see 5–10 feet of snow along its crest and western slopes over the next 10 days. Many mountain communities, like Tahoe City and Mammoth, should see 2–4+ feet of snow as well. Winter is back in California. https://t.co/rwq3QCGsWT - **scottmessinger** (2026-02-10): https://x.com/scottmessinger/status/2021054038243152309 - @kylemathews @thruflo I made a Cloudflare Worker implementation of Durable Streams: https://t.co/VmWmplpKwG It passes the conformance test suite (thanks for doing that!) and also includes an optional pub/sub layer when a client needs to listen to 100s of streams. - **btaylor** (2026-02-10): https://x.com/btaylor/status/2021041596293316848 - AI is your ghostwriter, but you are the author. I was speaking with my friend Arya about the complex dynamics software teams face now that most code is being generated by AI agents. There are very few natural bottlenecks to code “slop” - features over quality, and functionality over polish. I have made a point of banning speaking about AI as the author of code. Codex/Claude didn’t write that code, just like your table saw didn’t cut the wood - you did, with the help of AI. If we start speaking about AI that way, we remove the accountability of the engineer for the quality of the system they are producing. Arya put it succinctly in a way that I hadn’t heard before: AI is your ghostwriter, but you are the author. It’s your name on the cover of the book. I think the next year of software engineering will be building out the tools that enable all of us to be proud to sign our name to and take accountability for the increasingly complex systems we produce with our agents. - **southpolesteve** (2026-02-10): https://x.com/southpolesteve/status/2021036987009073447 - Trying out Codex desktop app and it is very good. But I miss sub agents. I had not realized how much I was leaning on them with OpenCode. Also I need remote support. `opencode serve` was a smart move. Terminal UIs get remote basically for free. - **ritakozlov** (2026-02-10): https://x.com/ritakozlov/status/2021028880228999257 - no better marketing than jane poking around. welcome back! @wongmjane what else are we shipping? i can’t even keep track anymore https://t.co/2Any1aUan4 - **wongmjane** (2026-02-10): https://x.com/wongmjane/status/2021023559028822431 - Cloudflare is working on Observability Alerts “Track your Workers’ metrics and get notified when they exceed defined thresholds” It does feel similar to Sentry Alerts… https://t.co/YUfc5E9MXL - **KeycardAI** (2026-02-09): https://x.com/KeycardAI/status/2020999175911981065 - MCP, agentic systems, and auth are moving fast 🤖 This Wednesday evening in SF, Keycard is co-hosting with @StainlessAPI, alongside @Cloudflare, @humanlayer_dev, and @temporalio, to talk MCP, auth, and where agent architectures are headed. Building with agents? Come join us. RSVP: https://t.co/1qryh29MgH - **eastdakota** (2026-02-09): https://x.com/eastdakota/status/2020989563993223555 - .@matt_levine remains a national treasure. https://t.co/yoqcODiLrZ - **dexhorthy** (2026-02-09): https://x.com/dexhorthy/status/2020985595946991710 - they're calling it "architecting agentic systems" i'm calling it "the roast of ian livingstone" see you wednesday https://t.co/cm4AjimjDV - **VegetablesReidy** (2026-02-09): https://x.com/VegetablesReidy/status/2020957658698907669 - @Cloudflare is just killing it right now. The platform is more delightful by the day! https://t.co/aDnmZaY9oP - **ConnorWaslo** (2026-02-09): https://x.com/ConnorWaslo/status/2020921635096887632 - we cut half of our product and it's so much better now. huge glow up and the makeover is just getting started. https://t.co/pazPpXAjAv - **RhysSullivan** (2026-02-09): https://x.com/RhysSullivan/status/2020917471944094037 - they built palantir for dogs and didn't call it datadog https://t.co/mWxwUPwiFn - **burcs** (2026-02-09): https://x.com/burcs/status/2020910817575633061 - working on a few different projects making them more agentic friendly... have been having a lot of success just stepping back and asking ai what it wants from the project, what's missing, and then having it implement those features results afterwards have been night and day - **StainlessAPI** (2026-02-09): https://x.com/StainlessAPI/status/2020909023529148667 - MCP, agentic systems, and auth 🤖 This Wednesday evening in SF with @KeycardLabs, @Cloudflare, @humanlayer_dev, and @temporalio. RSVP: https://t.co/S7DgeRyszf https://t.co/1bFzIeQnLZ - **ghostwriternr** (2026-02-09): https://x.com/ghostwriternr/status/2020862535075795274 - psa to vibe coding platforms: building interactive terminals for your users to work within a sandbox? start with these 10 lines of code and go-live with functioning, interactive access to a bash session. it's just. that. easy. https://t.co/X4FUIg38xv - **ghostwriternr** (2026-02-09): https://x.com/ghostwriternr/status/2020860573504721023 - why build on cloudflare sandboxes? <50 lines of code gives you: - interactive access to parallel bash sessions - built-in replays so you can reload and not lose state - durable objects to handle real-time sync all day @whoiskatrin and I had so much fun with this 🎉 so can you! https://t.co/gWI10d3Lug https://t.co/Z3EgVQxdAL - **whoiskatrin** (2026-02-09): https://x.com/whoiskatrin/status/2020857156078027159 - 🎈 we've just shipped full PTY support for Sandbox SDK you can now attach to a running sandbox and get a real interactive terminal out of the box support for creating terminals, replaying buffered output, resizing, connection retries go make cool stuff with it! - **zebassembly** (2026-02-09): https://x.com/zebassembly/status/2020849092121796942 - Ok what about git push agent <feature> instead of all this worktree slop - **LakeAustinBlvd** (2026-02-09): https://x.com/LakeAustinBlvd/status/2020827912786907622 - How it feels doing a morning lap around the @Cloudflare Lisbon office right now https://t.co/Rir3ijDDfb - **henrikruscon** (2026-02-09): https://x.com/henrikruscon/status/2020823971793350675 - Calendar in Alcove https://t.co/ASvtLZHquu - **_Evan_Boyle** (2026-02-09): https://x.com/_Evan_Boyle/status/2020718858873823447 - Unlimited tokens is the only perk that matters at your job. In the last 24 hours I used ~$5k worth of tokens to spike rewriting Copilot CLI in Bun + OpenTUI. No more ink flickering. - **snowmaker** (2026-02-09): https://x.com/snowmaker/status/2020709127065133562 - We're at an interesting stage of the evolution where the models are 1000x faster than I am at coding and 10x slower at using a web browser. - **gdb** (2026-02-09): https://x.com/gdb/status/2020706544011362423 - a fundamental difference in our respective outlooks on AI https://t.co/YAVcD7S4OO - **ankrgyl** (2026-02-09): https://x.com/ankrgyl/status/2020695633427087537 - over: database-wrapper software not over: llm-wrapper software - **samlambert** (2026-02-09): https://x.com/samlambert/status/2020686542981980169 - the openai ads were classy and inspiring. - **RohanNadkarni** (2026-02-09): https://x.com/RohanNadkarni/status/2020674164546261489 - This is incredibly cool — A source familiar with Bad Bunny’s performance tonight tells me that the couple in the halftime show got married for real. They invited Bad Bunny to their wedding and in turn he invited them to get married during his performance. - **LudwigAhgren** (2026-02-09): https://x.com/LudwigAhgren/status/2020673287609778527 - the bushes were people ???? https://t.co/GscO1tOHNY - **danshipper** (2026-02-09): https://x.com/danshipper/status/2020671009280467286 - BREAKING: At the Super Bowl halftime show, Bad Bunny says he prefers Opus 4.6 for vibe coding but turns to 5.3 Codex for gnarly engineering problems. “Opus pa’ vibear, Codex pa’ lo heavy” - he said - **Super70sSports** (2026-02-09): https://x.com/Super70sSports/status/2020669284507111438 - He’s already gained more yards than the Patriots. https://t.co/RvXmrvzLhS - **KentonVarda** (2026-02-08): https://x.com/KentonVarda/status/2020642903958405360 - Every test framework I've ever used that chooses to captures logs (stderr, console.log, etc.) has managed to make my life harder by doing so. Please, if you're making a testing framework, do not capture logs. - **EricTopol** (2026-02-08): https://x.com/EricTopol/status/2020566773092028759 - A review of the new trial results (below) along with all the other evidence and conclude it's time to move to AI-support for all mammography (with no cost to patients) https://t.co/hITOBNd7pg https://t.co/JDCUySM5bS https://t.co/TYQ36TiTbT - **dillon_mulroy** (2026-02-08): https://x.com/dillon_mulroy/status/2020537174626603182 - @irvinebroque i use codex after a bunch of context gathering, planning, and design. once it has that i’ll let it run for like an hour or two and come back to pretty successful work - **fkadev** (2026-02-08): https://x.com/fkadev/status/2020452350385635552 - 🔥 Introducing Textream, a macOS teleprompter that uses on-device speech recognition to highlight your script as you speak. Dynamic Island-style overlay, no cloud, works offline. Only 913kb 😍 → https://t.co/91QEaeC3L5 → brew install f/textream/textream - **ghostwriternr** (2026-02-08): https://x.com/ghostwriternr/status/2020443080185336287 - you don't have to wave and pray your docs are good when you've got a framework to follow, an agent who will and senses to tell when it's good seriously try this today https://t.co/SkmmD0bVMb - **threepointone** (2026-02-08): https://x.com/threepointone/status/2020424644319404388 - Free alpha for you, a thing you can do _today_. Point your docs at the Diátaxis framework https://t.co/N8LP9H21jD, ask your coding agent its opinions, and rewrite it. Learnt this from @ghostwriternr. It’s excellent. Been rewriting agents docs with this. https://t.co/FGsTzWfB1e - **theo** (2026-02-08): https://x.com/theo/status/2020357309537702227 - @gabriel1 I love music so much. I’ve never been able to work while listening to music. I use it as my reward after a long day. - **gdb** (2026-02-08): https://x.com/gdb/status/2020332743260008642 - video walkthrough of GPT-5.3 Codex: https://t.co/FNAfHPpMJv - **dillon_mulroy** (2026-02-08): https://x.com/dillon_mulroy/status/2020306614885183532 - i fear i've been sleeping on the codex models and they're drastically different in how they operate than opus - it's challenging some of my assumptions in how i've been thinking about working w/ ai - **theo** (2026-02-07): https://x.com/theo/status/2020279916760355142 - I've been using 5.3 Codex for 3 weeks. It's an incredible model. I've built so much stuff with it. Made a vid showing everything I love about it, as well as a few call-outs of things I hope OpenAI changes. https://t.co/aWzrRqRzXJ - **ericclemmons** (2026-02-07): https://x.com/ericclemmons/status/2020226853986836769 - I think worktrees are a short-term solution to what should be remote sandboxes. - **mattsimpsn** (2026-02-07): https://x.com/mattsimpsn/status/2020187708165955657 - first week at @cloudflare complete! ◇ visited the london office ◇ cracked team ◇ coffeeflare ◇ some incredible demos ◇ shipped some code the developer platform is only getting better from here https://t.co/N9R4lA8iky - **MarcSelwan** (2026-02-07): https://x.com/MarcSelwan/status/2020179011633377791 - Did you know that it's silly easy to create AI-assisted natural language search of your Apache Iceberg catalog (in this case, R2 Data Catalog)? This was a fun experiment - using a simple Cloudflare Worker: 1. Grab the latest snapshot(s) for your given tables 1.5. (Optional), you can easily filter out fields, convert to md files, or in my case, if your namespace/table prefixes are UUIDs, create a catalog-overview file to help search know how to map the UUIDs to human readable names. 2. Upload to R2 Then in the dash, click "Next" until the end when setting up AI Search. (Ok ok,you might have to enter the name of the prefix) Profit. The default embedding model seemed to be "good enough" and I'm sure you can get better results rolling your own via AI Gateway. - **jlongster** (2026-02-07): https://x.com/jlongster/status/2020168913653481886 - I implemented a decently complex feature using the same PLAN file with both Codex 5.3 and Claude 4.6, saved the implementation and ask Claude 4.6 to explain the differences Claude roasted its implementation: https://t.co/BwBBD6omKe - **steobrien** (2026-02-07): https://x.com/steobrien/status/2020154641388880175 - The moment we are now in https://t.co/iSg1MGYfRK - **ghostwriternr** (2026-02-07): https://x.com/ghostwriternr/status/2020142530222821644 - sandboxes 🤝 agents https://t.co/anF57U4hT8 - **jarredsumner** (2026-02-07): https://x.com/jarredsumner/status/2020138052929065391 - reading LLM-generated usage of node:fs is still very painful so much readFileSync(path).slice(0, max) so much if (existsSync(path)) statSync(path) so much readdirSync(path).map(_ => unlinkSync(_)) - **garrytan** (2026-02-07): https://x.com/garrytan/status/2020016721210536340 - I think about this every day https://t.co/1jAONL6ngh - **blader** (2026-02-07): https://x.com/blader/status/2019976271888019605 - i too found it very effective to give agents a "napkin" to write on as it works. it's a meaningfully different form of context than session history (lossy), or todos/plans (static) anyway, install this skill to give codex/claude a napkin to write on https://t.co/tr5iIf191O https://t.co/DO8jgP6zO8 - **JustJake** (2026-02-07): https://x.com/JustJake/status/2019925057116729824 - I don't understand why people don't do more content outside. Live a little, touch some grass. Fun, as always unfiltered/unplanned, discussion with Nikunj from FPV, talking about Railway and more https://t.co/82O7pQ2c15 - **nikunj** (2026-02-06): https://x.com/nikunj/status/2019916166370828435 - A Walk In The Park feat @JustJake 🚂 [00:00] 👋 [00:40] Is SaaS dead? AI's impact on software [03:58] Railway's $100M raise and vision [06:00] Origin story - from Uber to Railway [10:26] Engineering oath & business responsibility [13:39] Changed mind on weekends and PMs [16:15] Picking partners and fundraising [20:03] Final advice for founders 📽️ shot on iPhone and edited on @descript 📍 golden gate park - **koomen** (2026-02-06): https://x.com/koomen/status/2019911000334676262 - Things that seem important if you’re building software right now: 1. You can write code quickly now 2. Competitors can too 3. So can your users 4. If it takes more than a few clicks an agent should be able to do it 5. Users probably want to use their own agents - **XDevelopers** (2026-02-06): https://x.com/XDevelopers/status/2019881223666233717 - Officially launching X API Pay-Per-Use The core of X developers are indie builders, early stage products, startups, and hobbyists It’s time to open up our X API ecosystem and instill a new wave of next generation X apps We’re so back. https://t.co/FS66GuIYJD - **mattsimpsn** (2026-02-06): https://x.com/mattsimpsn/status/2019875581693669426 - new @cloudflare workers observability dashboard just shipped and it's packed with dx improvements! dms open - share what you're building or drop some feedback! https://t.co/gdKqAgcNcz - **thomas_ankcorn** (2026-02-06): https://x.com/thomas_ankcorn/status/2019860964187992215 - I like to joke around but the whole outerbase crew show the true power of showing up every day and shipping! The Cloudflare dash is not better because of some grand well thought out clever plans (these didn’t ship) It’s because this lot stubbornly made it better every day https://t.co/Nkwb45EHHN - **zebassembly** (2026-02-06): https://x.com/zebassembly/status/2019853935435976814 - @thomas_ankcorn absolutely cooked redesigning the new Workers Observability dashboard for the @Cloudflare developer platform. https://t.co/vmDhru20rM - **grinich** (2026-02-06): https://x.com/grinich/status/2019846500747288674 - > git push heroku master that single command changed everything heroku was the first time i realized you could combine utility and taste to build great developer tools. beauty, form, simplicity, and power in one package. it has inspired my entire career. sad to see it go. - **mschoening** (2026-02-06): https://x.com/mschoening/status/2019834729831494118 - RIP 🖤 Pour one out for `git push heroku master`. 🍻 Biggest acquisition fumble of all time? https://t.co/7y2hXvNKr7 - **amorriscode** (2026-02-06): https://x.com/amorriscode/status/2019833252794757299 - https://t.co/TbRluSfUYZ - **aarondfrancis** (2026-02-06): https://x.com/aarondfrancis/status/2019832589482356878 - I've been building Solo for several months, and it's out today! It's a desktop app that manages your entire dev stack. Add a project, Solo detects your processes, and you can start everything with one click. It's free! Try it today: https://t.co/804mFNwGV4 https://t.co/CXt9i9eb9Q - **ritakozlov** (2026-02-06): https://x.com/ritakozlov/status/2019830276776730840 - end of an era indeed... but the dream of heroku lives on at cloudflare! so much of what we do on the developer platform is inspired by that early magic that heroku was able to create if you haven't tried cloudflare's dev platform, this is your chance (now, with containers!) https://t.co/7uhWmGaxuE - **dbreunig** (2026-02-06): https://x.com/dbreunig/status/2019829245137338548 - A whenwords update, looking at more examples of spec driven development, including @vercel's just-bash, @pydantic's monty, and @AnthropicAI's Rust C-compiler: https://t.co/98R6qUjd7U - **YIMBYLAND** (2026-02-06): https://x.com/YIMBYLAND/status/2019802944506589280 - There are people with sub 80 IQs blasting 6,000lb behemoths through our neighborhoods and school zones. This should scare us all. https://t.co/zxIyN6h9qS https://t.co/UuEmWlkr0P - **threepointone** (2026-02-06): https://x.com/threepointone/status/2019800485331403017 - I called this incredibly hard https://t.co/quQ9A8GaoQ - **geoffreylitt** (2026-02-06): https://x.com/geoffreylitt/status/2019792700023672859 - "AI code review" should be about teaching people how the code works, not just verifying that it's correct We can now generate essays, diagrams, and explorable explanations on demand!! Why are we still reading raw code diffs as the primary UI? https://t.co/9jTS7XG8mq - **mschoening** (2026-02-06): https://x.com/mschoening/status/2019789558075060439 - I think we are vastly underprepared in terms of all infrastructure. Even the boring cloud infra we have relied on for compute 1.0 is not prepared for this growth. Not to mention the control planes for all of this, the data stores, the version control systems, the permission models etc. Everything has to change. - **garrytan** (2026-02-06): https://x.com/garrytan/status/2019787133360525518 - In the future we will wonder why we did this to ourselves. We will prompt and review a multi step plan, allow multiple agents argue for a few hundred milliseconds, read it and then just exit plan mode and if is wrong we will back it out and do it again the right way in hours. It’s two way doors everywhere! - **EastlondonDev** (2026-02-06): https://x.com/EastlondonDev/status/2019781616235671743 - Dillon is correct, I’m rocking cloudflare and datadog code mode mcps and Cursor and Claude Code are closing out alerts, finding bugs & performance bottlenecks and verifying changes on their own. It’s replaced hours I would spend each day clicking around looking at graphs in web dashboards - **threepointone** (2026-02-06): https://x.com/threepointone/status/2019716903120564606 - kenton varda fan club https://t.co/DQGmhPaPcG https://t.co/GIcR5JXAbg - **mikker** (2026-02-06): https://x.com/mikker/status/2019689268990595580 - I've built an entirely new, 100% native, extendable, themeable, modal cmd-space launcher for macOS. 🍣 Meet Tuna. https://t.co/9mFeIcBoUS https://t.co/8HVcqNmIj8 - **samuelcolvin** (2026-02-06): https://x.com/samuelcolvin/status/2019604402399768721 - Fuck it, a bit early but here goes: Monty: a new python implementation, from scratch, in rust, for LLMs to run code without host access. Startup time measured in single digit microseconds, not seconds. @mitsuhiko here's another sandbox/not-sandbox to be snarky about 😜 Thanks @threepointone @dsp_ (inadvertently) for the idea. https://t.co/UuCYneMQ9j - **boristane** (2026-02-06): https://x.com/boristane/status/2019589856410890251 - been here 1 month and every day feels just like this https://t.co/AcUGY65UYo - **gdb** (2026-02-06): https://x.com/gdb/status/2019566641491963946 - Software development is undergoing a renaissance in front of our eyes. If you haven't used the tools recently, you likely are underestimating what you're missing. Since December, there's been a step function improvement in what tools like Codex can do. Some great engineers at OpenAI yesterday told me that their job has fundamentally changed since December. Prior to then, they could use Codex for unit tests; now it writes essentially all the code and does a great deal of their operations and debugging. Not everyone has yet made that leap, but it's usually because of factors besides the capability of the model. Every company faces the same opportunity now, and navigating it well — just like with cloud computing or the Internet — requires careful thought. This post shares how OpenAI is currently approaching retooling our teams towards agentic software development. We're still learning and iterating, but here's how we're thinking about it right now: As a first step, by March 31st, we're aiming that: (1) For any technical task, the tool of first resort for humans is interacting with an agent rather than using an editor or terminal. (2) The default way humans utilize agents is explicitly evaluated as safe, but also productive enough that most workflows do not need additional permissions. In order to get there, here's what we recommended to the team a few weeks ago: 1. Take the time to try out the tools. The tools do sell themselves — many people have had amazing experiences with 5.2 in Codex, after having churned from codex web a few months ago. But many people are also so busy they haven't had a chance to try Codex yet or got stuck thinking "is there any way it could do X" rather than just trying. - Designate an "agents captain" for your team — the primary person responsible for thinking about how agents can be brought into the teams' workflow. - Share experiences or questions in a few designated internal channels - Take a day for a company-wide Codex hackathon 2. Create skills and AGENTS[.md]. - Create and maintain an AGENTS[.md] for any project you work on; update the AGENTS[.md] whenever the agent does something wrong or struggles with a task. - Write skills for anything that you get Codex to do, and commit it to the skills directory in a shared repository 3. Inventory and make accessible any internal tools. - Maintain a list of tools that your team relies on, and make sure someone takes point on making it agent-accessible (such as via a CLI or MCP server). 4. Structure codebases to be agent-first. With the models changing so fast, this is still somewhat untrodden ground, and will require some exploration. - Write tests which are quick to run, and create high-quality interfaces between components. 5. Say no to slop. Managing AI generated code at scale is an emerging problem, and will require new processes and conventions to keep code quality high - Ensure that some human is accountable for any code that gets merged. As a code reviewer, maintain at least the same bar as you would for human-written code, and make sure the author understands what they're submitting. 6. Work on basic infra. There's a lot of room for everyone to build basic infrastructure, which can be guided by internal user feedback. The core tools are getting a lot better and more usable, but there's a lot of infrastructure that currently go around the tools, such as observability, tracking not just the committed code but the agent trajectories that led to them, and central management of the tools that agents are able to use. Overall, adopting tools like Codex is not just a technical but also a deep cultural change, with a lot of downstream implications to figure out. We encourage every manager to drive this with their team, and to think through other action items — for example, per item 5 above, what else can prevent a lot of "functionally-correct but poorly-maintainable code" from creeping into codebases. - **thdxr** (2026-02-05): https://x.com/thdxr/status/2019559378400211277 - i've been a manager, i'm partially a manager now and let me tell you something it's absolutely nothing like managing a bunch of ai agents - **jainarvind** (2026-02-05): https://x.com/jainarvind/status/2019553277571190821 - https://t.co/6uiXmh6CsI - **ritakozlov** (2026-02-05): https://x.com/ritakozlov/status/2019488964097224838 - opus 4.6 available on @cloudflaredev ai gateway AND @replicate days like this are a good reminder of why the ability to switch models via just a line of code is so powerful https://t.co/ZOhqDOmtCa - **southpolesteve** (2026-02-05): https://x.com/southpolesteve/status/2019473533517123718 - Free idea. Please someone vibecode this https://t.co/uJzTLrxuJ3 - **threepointone** (2026-02-05): https://x.com/threepointone/status/2019462619128607110 - @samuelcolvin @github @NotionHQ @southpolesteve was just complaining to me about this Google docs but for markdown with github/gist backend - **thdxr** (2026-02-05): https://x.com/thdxr/status/2019452227438489897 - i think a lot more about how things would be with 10x faster LLMs than 10x smarter LLMs - **zachweinberg** (2026-02-05): https://x.com/zachweinberg/status/2019434244657885246 - By far the #1 trait I've noticed in great startup hires is the quality of the questions they ask about the business during an interview. How much do they dig in, do they ask clarifying questions when my explanation isn't good enough, do they push on second and third order problems etc. - **elithrar** (2026-02-05): https://x.com/elithrar/status/2019417459368030265 - we've just added a new agent skill for the @cloudflare Sandbox SDK to our skills catalog: ◆ npx skills add cloudflare/skills │ ◼ sandbox-sdk (Build sandboxed applications for secure code execution.) https://t.co/eD3IT4EIkz - **thdxr** (2026-02-05): https://x.com/thdxr/status/2019402921536843787 - tired of the fear narrative pushed by ai companies i try to highlight ways trying hard and caring still matter - i think it means a lot coming from people in the space others who are pro ai and pro human worth following: @martin_casado @pvncher @leerob (cursor in general) @0xblacklight @realmcore_ @charlieholtz - **dickson_tsai** (2026-02-05): https://x.com/dickson_tsai/status/2019314348632555651 - In Claude Code v2.1.30, we introduced /debug, a built-in skill for Claude to read your session's debug logs and troubleshoot your session. Great for chatting through issues like "/debug why didn't my hook trigger?" or "/debug why did my tool call fail?" How did /debug come about? Last week, I was observing a group of users onboarding onto Claude Code. I saw that when something unexpected occurred, it wasn't easy for us to figure out the cause from the limited TUI display. Meanwhile, it was a chore to locate the logs. What if we simply gave Claude access to it from within the TUI? That way, it can use its tools to scan the logs and/or pull in the claude-code-guide subagent. The user can also /rewind when they're done debugging. This feature is only as useful as what's in the debug logs. We're open to suggestions about what else we should add to them! Then, /debug should get better over time. - **lessin** (2026-02-05): https://x.com/lessin/status/2019263818560598133 - https://t.co/CP12TXx4OC - **thdxr** (2026-02-05): https://x.com/thdxr/status/2019260077228880103 - is it possible to make software that you're proud of - **theo** (2026-02-05): https://x.com/theo/status/2019259106960527443 - Claude Code is so 2025 - **arampell** (2026-02-05): https://x.com/arampell/status/2019237300337824010 - Anyone who’s written software knows that the last 5% can take 95% of the time Edge cases, conflicts, weird environments, you name it And for many areas of work, 95% right is still 100% wrong AI is a remarkable tool and a huge productivity gain. Laws of physics still apply. - **eric_seufert** (2026-02-05): https://x.com/eric_seufert/status/2019231107321827776 - This Anthropic ad is simply obnoxious. Anthropic generates most of its revenue from its enterprise business, so it can afford not to maximize its consumer revenue opportunity. But the sanctimonious moralizing here presents advertising as a cynical business model choice. It is neither cynical nor a choice; the freemium, digital advertising-supported model is the only repeatedly proven pathway for a consumer technology product to reach humanity scale. OpenAI’s advertising revenue will enable it to offer access to larger models to free tier users. To the extent Anthropic thinks chatbots powered by frontier models are valuable to society, it should aspire to provide access to as many people as possible. Ads allow for that to an extent that no other consumer technology business model, demonstrably, can. This kind of condescending attitude toward digital advertising is economic chauvinism, and it represents a desire for technological gatekeeping. And the idea that ads will necessarily influence chatbot content is overstated, borderline anachronistic. It ignores the functional form of most modern ad platforms, which principally use behavioral history, not context, to personalize advertising. Ads in chatbots shouldn’t be seen as a perfect replacement for Search ads; an entirely new category of display advertising emerged and flourished after Search. - **OpenAIDevs** (2026-02-05): https://x.com/OpenAIDevs/status/2019221475849564657 - ⚙️ Inside the Codex harness All Codex surfaces, including the Codex app, the Codex CLI, the Codex web app, and IDE integrations — like @Code, Xcode, and @JetBrains — are powered by the same Codex harness under the hood. We're sharing details on the Codex App Server, a JSON-RPC protocol that exposes this harness for integration. https://t.co/NdqI98cZpc - **JustJake** (2026-02-05): https://x.com/JustJake/status/2019200687914250562 - The best advice is listened to diligently, intentionally digested, and then promptly discarded Anything that sticks forms the basis for you. Anything that does not, probably isn’t right for you at this time https://t.co/yfcWGlgJXz - **mxstbr** (2026-02-04): https://x.com/mxstbr/status/2019186112800125138 - Second big ship at OpenAI 🚀 https://t.co/b7BqRbxB0H - **MarcSelwan** (2026-02-04): https://x.com/MarcSelwan/status/2019152722541605300 - Turns out, this works beautifully with R2 Data Catalog (Apache Iceberg). if you have Iceberg clients writing cross-region, you can quite the latency boost with a simple: npx wrangler r2 bucket local-uploads enable <R2 Data Catalog Name> Nice work @aninibread and team! https://t.co/kBea8afHy0 https://t.co/vEKmSU59QT - **nicbstme** (2026-02-04): https://x.com/nicbstme/status/2019149771706102022 - https://t.co/7BHqiRygfP - **samgoodwin89** (2026-02-04): https://x.com/samgoodwin89/status/2019144500711682230 - I asked the Cloudflare team to add OAuth scopes for AI Search and not only did they add them, they also submitted a PR fixing alchemy’s AI search resources. So awesome. Cloudflare team gets better every year. - **ankrgyl** (2026-02-04): https://x.com/ankrgyl/status/2019138618376483052 - AI-powered coding bloats the illusion of work more significantly than it increases the output itself. Of course, it does (or can) increase output. But it's easy to be tricked by the illusion of work, and we have to learn to un-train our intuition that code -> product output. - **ankrgyl** (2026-02-04): https://x.com/ankrgyl/status/2019132517836886291 - everyone should do this. open up codex, paste this tweet, and have it audit your code. https://t.co/zr4t2Y88TU https://t.co/mcoIJHNRhp - **boristane** (2026-02-04): https://x.com/boristane/status/2019126973000478761 - it's going to be so funny when anthropic inevitably puts ads in claude - **destraynor** (2026-02-04): https://x.com/destraynor/status/2019091578451566736 - @saranormous Yep , it's equal parts "we need to go beyond AGI to mimic our refund policy" + "do you integrate with Microsoft Dynamics" + "on prem too, if possible" - **miacmal** (2026-02-04): https://x.com/miacmal/status/2019090906771517712 - what’s new in workflows🧵 @CloudflareDev (1) you can now visualize your workflows in the cloudflare dashboard (2) build durable background agents with workflows support in agents SDK v0.3.7 - **ritakozlov** (2026-02-04): https://x.com/ritakozlov/status/2019088383679619334 - no better place to build (and now visualize!) agent workflows than on @cloudflaredev → https://t.co/MGGvrEe9Q6 - **BraydenWilmoth** (2026-02-04): https://x.com/BraydenWilmoth/status/2019078147992351223 - At Cloudflare we have been building a new component library on top of @base_ui – we call it Kumo. It's an opinionated style layer on top of accessible components with some additional tools, such as: - Pre-styled components - Figma plugin (code -> design) - Streaming UI support - AI-ready component registry - CLI Why are we sharing this? We love to give back to the internet. We don't expect this to be a library that everyone adopts and uses in all of their projects. Maybe you like it and want to use it & that's great. But we can share our tooling and our learnings with the community to make developing beautiful UI's easier. If you have been looking at the Cloudflare dashboard UI for the past few months you have probably noticed this new component library in action. While we have a long way to go to make it perfect.... we're happy with how it's coming along! Huge shoutout to all the contributors who have helped build this internally. Wouldn't be where it is without all the help :) @JonnieLappen @mattrothenberg @superhighfives @nandafyi @hkfoster @invisal89 @RozenMD @bradgarropy + Rob & many more - **CloudflareDev** (2026-02-04): https://x.com/CloudflareDev/status/2019077718437220795 - SF, come get coffee! Join us for Coffeeflare at @Sightglass on 20th St, starting at 8am PT ☕️ https://t.co/hgmBU38f9Y - **michellechen** (2026-02-04): https://x.com/michellechen/status/2019064756036878658 - hmmm mm https://t.co/h4uYQabw6P - **thdxr** (2026-02-04): https://x.com/thdxr/status/2019058747847459206 - things don't have to be great to work most food is bad, most software feels terrible when you come across something that feels great it's because they did more than they had to and rejected the tradeoffs everyone told them were real - **saranormous** (2026-02-04): https://x.com/saranormous/status/2019057459529449672 - In order to have a sense of AI adoption right now, you really need to have one foot at the bleeding edge with the kids Wispr-ing through a mic at Devin all day and one foot in the 99% enterprise world where they’re still debating the merits of leaving “bundled-MSFT-ELA-Copilot” - **zebassembly** (2026-02-04): https://x.com/zebassembly/status/2019042785819763116 - Super whisper is so good why have I not been using this - **jarredsumner** (2026-02-04): https://x.com/jarredsumner/status/2019006416791740596 - @adamdotdev This “adult in the room” framing is pretty rude to the Claude Code team that built a product hitting $1B run-rate revenue faster than probably anything in history. Bun made like $2.50 total (stickers). Engineering is relative to time & tradeoffs & they made fantastic tradeoffs - **garrytan** (2026-02-04): https://x.com/garrytan/status/2018945166833524962 - Nearly 40% of Stanford undergrads now claim a disability, vs 4% at community colleges. The problem isn’t 18-year-olds gaming incentives. It’s elites building a world where honesty is for suckers. Cluely is not everything. Bring back integrity. https://t.co/qszPvDe0Fx - **dwallacewells** (2026-02-04): https://x.com/dwallacewells/status/2018881806393852136 - Simply giving immunotherapy earlier in the day cut cancer mortality roughly in half, https://t.co/MoFFtvpxuU - **tjcages** (2026-02-04): https://x.com/tjcages/status/2018859145366303158 - important for AI-friendly sites to have auto-generating markdown for every page route llms.txt acts as a sitemap, pointing agents to each .md path for seamless navigation AI-friendly by default https://t.co/Dd8m6C1aA2 - **theo** (2026-02-04): https://x.com/theo/status/2018841708705038405 - Wait what https://t.co/bGKS9EDZHs - **OpenAIDevs** (2026-02-04): https://x.com/OpenAIDevs/status/2018838297221726482 - GPT-5.2 and GPT-5.2-Codex are now 40% faster. We have optimized our inference stack for all API customers. Same model. Same weights. Lower latency. - **theo** (2026-02-03): https://x.com/theo/status/2018819504252608710 - Good to see the industry finally standardizing. Products that moved to `.agents/skills` so far: - Codex - OpenCode - Copilot - Cursor - Not Claude Code https://t.co/rU1hVOq3L9 - **swyx** (2026-02-03): https://x.com/swyx/status/2018816288475537741 - anyone that believes openai has given up on fundamental research and is simply productmaxxing has not internalized this chart my sweet summer child; Sora was like a 5 person side project. Deep research. Even initially Codex was 2 side projects (scaled up now ofc). we always always only see the tip of the iceberg. - **kathyyliao** (2026-02-03): https://x.com/kathyyliao/status/2018801634600718792 - Browser Rendering makes it easy to generate PDFs from html 👇 https://t.co/jPCfg4m5zI - **ankrgyl** (2026-02-03): https://x.com/ankrgyl/status/2018784613699923988 - weird moment, harness speed now matters as much as model speed. old school optimization is cool again for a bit. - **aninibread** (2026-02-03): https://x.com/aninibread/status/2018769961242485208 - 1 interactive diagram = 1,000 words. Did you catch this in the Local Uploads docs? 👀 https://t.co/4rB9fo6UKq - **ritakozlov** (2026-02-03): https://x.com/ritakozlov/status/2018707317030604818 - ICYMI: deploy to straight to @cloudflare from the @openai codex app https://t.co/7uhdYykiqI https://t.co/BMqQWe3puX - **ritakozlov** (2026-02-03): https://x.com/ritakozlov/status/2018690690096546274 - the best part of getting to build our products on our developer platform is getting to benefit from it this is what it looks like when we basically started offloading async work to cloudflare queues R2 uploads became 75% faster https://t.co/FhSem6HnAy - **aninibread** (2026-02-03): https://x.com/aninibread/status/2018687962964255007 - Today we launched R2 Local Uploads to improve upload performance. Here's how it works 🧵 https://t.co/LufB5nPsiD - **threepointone** (2026-02-03): https://x.com/threepointone/status/2018665683299811624 - also landed in the repo, and working on publishing publicly as soon as we're happy with it: an interactive playground for agent sdk, for _every_ feature and pattern. Coming Soon ™ https://t.co/0MnS5Nq2TZ - **CloudflareDev** (2026-02-03): https://x.com/CloudflareDev/status/2018664391403774103 - Big update to the Agents SDK. Run `npm i agents@latest` to get started. https://t.co/J7MVZuaOc7 - **threepointone** (2026-02-03): https://x.com/threepointone/status/2018650110499180954 - I'm going to make these docs so amazing for all your coding agents to swallow that ideally you'll never ever ever have to read them 😆 https://t.co/NkIUDAQmBU - **threepointone** (2026-02-03): https://x.com/threepointone/status/2018649649310032302 - Workflows in Agents! This took a while, but I'm so happy it's out there now, and it's better than a bunch of options out there (imho hehe). Live real time tracking of workflows, inbuilt into every agent. Have a look at the docs here https://t.co/EGLdrwgsJx (landing in main workers docs this week) - **whoiskatrin** (2026-02-03): https://x.com/whoiskatrin/status/2018647628980875473 - we just shipped one of our biggest Agents SDK updates in a while! 🎈 - 15+ new docs with more examples - interval scheduling with scheduleEvery() - custom routing - callable timeouts - state validation hooks - MCP options api and so much more! go build cool stuff! - **lucataco** (2026-02-03): https://x.com/lucataco/status/2018504484884787447 - it's never been easier to deploy to @Cloudflare 🔥 one of the top skills in Codex out of the box! https://t.co/ciUZugwMO6 https://t.co/ZfCNO3pMOT - **theo** (2026-02-03): https://x.com/theo/status/2018488771453587796 - I really don’t see Google catching up any time soon. They’ve baked so much intelligence into their models and they are still so unpleasant to use. Beyond the model, the software matters more and more, and they are a decade behind there. - **ctatedev** (2026-02-03): https://x.com/ctatedev/status/2018477472564887948 - TIL the Vercel CLI has a built-in 𝚌𝚞𝚛𝚕 for hitting endpoints behind deployment protection 𝚟𝚎𝚛𝚌𝚎𝚕 𝚌𝚞𝚛𝚕 /𝚊𝚙𝚒/𝚑𝚎𝚊𝚕𝚝𝚑 I only found out bc an agent automatically reached for it after realizing it was about to run a command against a Preview deployment 🤯 - **OpenAIDevs** (2026-02-02): https://x.com/OpenAIDevs/status/2018460509922857022 - Here's how our team uses the Codex app 👇 @edbayes shows about using the @figma skill to implement designs with 1:1 visual parity, including all CSS classes and styling automatically https://t.co/nMZJHtKnst - **zeeg** (2026-02-02): https://x.com/zeeg/status/2018438903238414676 - I wrote up a bit more on the never ending debate on MCP vs CLI, and I find it hard to believe anyone rational can read this and tell me one replaces the other. https://t.co/8sJwandj9H warning: its mostly me annoyed that this discourse continues to exist https://t.co/pXQFtOLtw2 - **sama** (2026-02-02): https://x.com/sama/status/2018437537103269909 - To celebrate the launch of the Codex app, we doubled all rate limits for paid plans for 2 months! And added access for free/go. - **kevinweil** (2026-02-02): https://x.com/kevinweil/status/2018418964708331757 - I've gotten to the point where if I'm sitting in a meeting and I don't have a Codex prompt running, I feel anxious for "wasting" an hour I could have been making parallel progress. It's happened super quickly and totally changed how I think. - **ParthGujare_** (2026-02-02): https://x.com/ParthGujare_/status/2018418361001025709 - we've been quietly building our own 0 -> 1 revenue stack internally at @tryramp - Ramp Revenue. our team’s mandate is simple - help our sales team win, drive more pipeline, and build the most efficient gtm org in the world. it's powered by our customer data platform (processes millions of records of internal, external, and crm data daily) + unified action layer with agents embedded directly in their workflows. This means that sellers at Ramp don’t have to worry about switching between dozens of systems just to figure out who they should reach out to or what they should say. >80% of sales workflows are now powered by Ramp Revenue. Sharing a small preview of what we've been building - **embirico** (2026-02-02): https://x.com/embirico/status/2018415923930206718 - 📣 Open call to agent builders: Let's read agent skills from `.agents/skills`, so people don't have to manage separate folders per agent. Today we pulled the trigger for Codex to read `.agents/skills`. Goal is to deprecate `.codex/skills`. Pls like/tag/RT for momentum. - **sama** (2026-02-02): https://x.com/sama/status/2018414858015039504 - Codex app is out for mac! I am surprised by how much I love it; it is a bigger step forward than I imagined. Lots more to come. - **deredleritt3r** (2026-02-02): https://x.com/deredleritt3r/status/2018408770808807785 - Scheduled tasks in Codex. This is huge. For example, I should now easily be able to ask Codex to run a daily search for newly introduced legislation. Or new publicly available agreements of a certain type. Or new guidance issued by regulators. So easy, even a lawyer can do it. https://t.co/sv64Ow1Akh - **embirico** (2026-02-02): https://x.com/embirico/status/2018389334584692778 - I'm Alexander and I'm part of the awesome Codex team at @OpenAI. We just launched an app! The app makes it easy to run agents in parallel, across projects. It has great support for skills, and setting up background automations on worktrees. Can't wait to hear what you think! https://t.co/uOiqP3nT96 - **grinich** (2026-02-02): https://x.com/grinich/status/2018376482209288351 - This weekend I built "WorkTV" It's video library for your team’s meeting recordings. Syncs recordings from Zoom and Gong. Browse, search, clip, and share with full transcript support and AI-powered summaries. https://t.co/lAFSh2Dcvn - **antirez** (2026-02-02): https://x.com/antirez/status/2018363173754073342 - Anthropic may be near to release a new model, however, as things are today, after weeks of testing, I'm totally sure Codex with GPT 5.2-xhigh is definitely a model capable of handling way more complex tasks. - **EricTopol** (2026-02-02): https://x.com/EricTopol/status/2018356155362488513 - The time of day for cancer immunotherapy is associated with major outcomes. Early is better. Results from a randomized trial of lung cancer, backs up the importance of our circadian rhythm and immune system https://t.co/bHqUZ3U83O https://t.co/beiN77kp1T - **JaredSleeper** (2026-02-02): https://x.com/JaredSleeper/status/2018348063283138703 - One thing that is abundantly clear is that traditional software growth investing (whether private or public) is dead. Metrics have perhaps never mattered less in software investing than they do today. The assumption used to be that the world moved fairly slowly/predictably, and so you could extrapolate trends in one place and assume they more or less continued. All of the old techniques (LTV:CAC math, unit economics, etc.) assumed this stability. That stability is a bad assumption today and it is not how assets are priced- the market is scrambling to sort out the new technical equilibrium and which companies have a place in it/are beneficiaries. I still think many incumbents are well-positioned, but importantly, their near-term metrics don't tell you anything about that at all. More than ever, there is just no substitute for truly, deeply understanding the world and how products fit in/are built/etc. and staying at the cutting edge of understanding AI adoption. You need to be more technical, deeper in the product, understand narratives better, understand teams and execution etc. One could argue even investing in public SaaS today is more analytically akin to venture investing than traditional public markets investing. - **zachkrall** (2026-02-02): https://x.com/zachkrall/status/2018347272791085203 - https://t.co/YIkPY0Lv1C - **zachkrall** (2026-02-02): https://x.com/zachkrall/status/2018338627563212832 - the terminal is quietly becoming one of the most interesting surfaces in interaction design i've been experimenting with what a personal website feels like when it lives in the terminal instead of a browser https://t.co/E3cySMyFRC - **elithrar** (2026-02-02): https://x.com/elithrar/status/2018323896273981588 - this is undoubtedly where things are headed (and many places have been practicing for some time). if you’re attached to a singular title vs focused on solving problems then now might be a good time to reconsider. https://t.co/XT7PImQjDt - **WillManidis** (2026-02-02): https://x.com/WillManidis/status/2018318844247347308 - https://t.co/qNAVuG3xsz - **thsottiaux** (2026-02-02): https://x.com/thsottiaux/status/2018258151603388639 - Codex now pretty much builds itself, with the help and supervision of a great team. The bottleneck has shifted to being how fast we can help and supervise the outcome. - **AdamNMayer** (2026-02-02): https://x.com/AdamNMayer/status/2018139108359733252 - Nighttime at the Transamerica Pyramid Redwood Grove is a vibe. https://t.co/v5gqBhFy7w - **matt_slotnick** (2026-02-02): https://x.com/matt_slotnick/status/2018137133769433439 - prevent defense never works. and incumbents are playing prevent defense - **matt_slotnick** (2026-02-02): https://x.com/matt_slotnick/status/2018137023471853658 - this is extremely hard for incumbents, because it involves de-centering the very thing you've been marketing for the last decade. but accepting that the new layer is new and winning it on its own merits is how you reinforce the dominance of the layers below - **matt_slotnick** (2026-02-02): https://x.com/matt_slotnick/status/2018136679232774654 - software has a perception problem. investors want to see claude code or lovable type growth, but forcing customers through existing interfaces all but ensures you won't see that. owning the core is a huge advantage, but you need to build the new layer outside in, not inside out - **sriramk** (2026-02-01): https://x.com/sriramk/status/2018111651996971490 - I have come to see the AI ecosystem in the US is a multi layered loop in the @kevinakwok sense. -> build out data centers ( with energy generation) -> populate the data centers with GPUs / TPUs / ASICs. Each of these is a little ecosystem - networking , memory, software, etc, each having its own ecosystem. -> resource allocate those data centers to research / training and inference of frontier models. -> inferencing of those models / applications on top drive usage, improvements to models and further financing. When these loops align you see exponential growth. Challenges arise when the cadence of these loops are with different timelines. I’ve had great success using this visual to explain a lot of the current state of AI to people. - **theo** (2026-02-01): https://x.com/theo/status/2018091358251372601 - https://t.co/zugEHNkncO - **zeeg** (2026-02-01): https://x.com/zeeg/status/2018087611659952268 - Whoever makes a coding harness that properly allows isolation of toolsets (MCP required!) to different subagents will be the people that move us forward. - **ankrgyl** (2026-02-01): https://x.com/ankrgyl/status/2018083485945344275 - pro tip. work on ~4 projects at once use opencode (kimi k2.5), codex, claude code, and cursor (one in each workspace) it's super valuable to feel the trade offs - **sberens** (2026-02-01): https://x.com/sberens/status/2018057486922834416 - https://t.co/PfxmFm0D89 - **thdxr** (2026-02-01): https://x.com/thdxr/status/2018051250785272127 - "who cares if it's slop it works code doesn't matter" isn't a good way to sell ai im producing cleaner codebases than ever because of ai now i can actually refactor everything when we find a better way of doing something before shitty code just lingered forever - **lessin** (2026-02-01): https://x.com/lessin/status/2018039906581500098 - I hesitate to point this out ... but if California were smart they wouldn't add a tax for billionaire residents -- they would charge a huge tax on landing gulfstreams in California. Way more aligned and effective. - **alexatallah** (2026-02-01): https://x.com/alexatallah/status/2018031924497838441 - .@infisical has been a great way for our team to centralize API key management Helps you avoid feeling "stuck" to a cloud provider https://t.co/sgaGb69XMN - **nbaschez** (2026-02-01): https://x.com/nbaschez/status/2018027072720130090 - Single biggest improvement to your https://t.co/KUZC0h59Pa / https://t.co/LTwkykSOrf: "When I report a bug, don't start by trying to fix it. Instead, start by writing a test that reproduces the bug. Then, have subagents try to fix the bug and prove it with a passing test." - **gokulr** (2026-02-01): https://x.com/gokulr/status/2018007131904241874 - PREDICTION: THE CPO ROLE, AS WE KNOW IT, WILL VANISH IN FIVE YEARS At young AI native companies, the traditional PM role is on the wane, replaced with a product builder archetype that’s a combination of Product, Design and Engineering. These companies will never hire a CPO. A separate product leader leads to too much cognitive dissonance when the IC roles doing the actual work are blending, extra overhead and imposes an unnecessary coordination tax on the product development organization. Five years from now, these companies will be the leaders and set the cultural tone for the next generation, so my prediction is that all tech companies will stop hiring for the CPO role in five years. There will be a singular product development leader at each org. Ironically, this new role might still be called the CPO, except they will run the entire product development org. CPTO is far too unwieldy of a title and only exists today to alleviate confusion. Career implication: early / mid career product leaders need to stop aspiring to become CPOs. instead, you need to develop a panoply of product development skills across all three disciplines (+ analytics), be able to fluidly navigate the roles, and become a product builder, period. Farewell, CPO! It was a good 15-20 year run for this role in tech. But like everything else, it’s time to evolve. - **mitsuhiko** (2026-02-01): https://x.com/mitsuhiko/status/2017933039997780475 - I wish people would build more useful open source projects and fewer open source products. Everything now has a flashy website and competes for attention, and when it doesn’t get traction the main dev bounces. - **thdxr** (2026-02-01): https://x.com/thdxr/status/2017822494367871348 - i'm pretty sure anthropic will one day make a model that's not available in the api and locked to claude code and i bet they'll say it's for "safety" reasons - **theo** (2026-02-01): https://x.com/theo/status/2017797919542538742 - Claude Code makes my M4 Macbook feel like an Intel Macbook https://t.co/ZGLwOiFrqV - **thdxr** (2026-02-01): https://x.com/thdxr/status/2017783221174321660 - it's worth having a bit of patience when a new model drops just in the last 24 hours can see how tps, ttfb, and errors have been drastically improving for kimi k2.5 in zen https://t.co/XvMxPCJiH6 - **thdxr** (2026-02-01): https://x.com/thdxr/status/2017756481559339221 - exactly the same i use gpt for big difficult things and kimi k2.5 for everything else claude models are in an awkward zone where they're smarter, but not enough and too expensive and slow https://t.co/p9QHRlZC3p - **jyobo10** (2026-01-31): https://x.com/jyobo10/status/2017732209252438021 - https://t.co/t8aboMkpaN - **theo** (2026-01-31): https://x.com/theo/status/2017725704386056214 - I was wrong. Claude Code's performance is unacceptable. Whenever I open it, my laptop's fans start roaring and my battery goes to shit. No issues with Codex or OpenCode. https://t.co/3hZMFdvK99 https://t.co/H6UwtGkFpk - **thdxr** (2026-01-31): https://x.com/thdxr/status/2017691649384620057 - finally got around to setting up an always on opencode server so i can run sessions on any device from anywhere takes a few minutes - showed it off here https://t.co/wIVGqlTbpQ - **thdxr** (2026-01-31): https://x.com/thdxr/status/2017666712603091378 - ai is such a magnifier of judgement that it's going to get really obvious who knows what they're doing and who doesn't - **akothari** (2026-01-31): https://x.com/akothari/status/2017663493819011271 - don't think market has yet priced in kimi k2.5 - **garrytan** (2026-01-31): https://x.com/garrytan/status/2017650291282526645 - Calling the most reasonable center-left people "right wing pundits" is what... frankly a lot of people do. They do that to me all the time. When you see it, you should say something. That's not useful, and it will doom the Democratic party to a decade of failure. If you really want to fight Trump, Democrats must be competent instead of clout chasing extremists. - **lessin** (2026-01-31): https://x.com/lessin/status/2017636447646601460 - HumansOnly social network is next step.... should have been built ages ago.. molt is last straw. - **scottbelsky** (2026-01-31): https://x.com/scottbelsky/status/2017626034644934865 - https://t.co/VjDJe5Zo2R - **yuris** (2026-01-31): https://x.com/yuris/status/2017605454378242154 - I think this is a fair hypothesis but there’s no reason to believe it will take startups 1-2 years instead of day/weeks to build migration tools. IMO migration tools might be some of the easier things to vibecode. https://t.co/685HaB2RjK - **robin_j_brooks** (2026-01-31): https://x.com/robin_j_brooks/status/2017578412294946921 - Yesterday's rise in the Dollar and the big crash in precious metals is fooling many into thinking that markets see Warsh as a hawk. They don't. Here's the 2-year yield, which fell steadily after the 6:48 am Truth Social post. Warsh will cut early and hard. https://t.co/3Xhp9PxFjW https://t.co/5UhYL4fw05 - **theo** (2026-01-31): https://x.com/theo/status/2017576945299689580 - Configuring VS Code via Claude Code. The magic still hasn't worn off. https://t.co/BuXMtXIQbN - **Baconbrix** (2026-01-31): https://x.com/Baconbrix/status/2017464112658452928 - Expo Router is the most beautiful React library https://t.co/bfrWxNXWo4 - **thomasgauvin** (2026-01-31): https://x.com/thomasgauvin/status/2017403232692084817 - "small" because this is a parameter change "big" because we're giving you more control over @Cloudflare Workers KV's consistency, and it's a first step in that direction https://t.co/uTXB4bgq5N https://t.co/oXdruXRCFl - **dflieb** (2026-01-31): https://x.com/dflieb/status/2017402890542059675 - “Well you see, honey, there were some employees at OpenAI who worried about the AI taking over, so they built a thing called Claude. Then another guy made a thing called Clawdbot, and Claude made him rename it to Moltbot. Then someone made a website called Moltbook for the Moltbots to talk to each other, tho it’s more like Moltreddit. And yeah anyway it’s funny or maybe really bad who knows lol” - **jeffreyhuber** (2026-01-30): https://x.com/jeffreyhuber/status/2017376561314599270 - @irvinebroque @Sightglass will try to come by! we are a block away - **matt_slotnick** (2026-01-30): https://x.com/matt_slotnick/status/2017365675816452545 - it's weird to me that every single release from a lab is made out to be a whole big thing that changes the future of work... but of all of them, this one seems potentially the most consequential and i've seen literally no talk about it https://t.co/Fob9kQaYPo - **ritakozlov** (2026-01-30): https://x.com/ritakozlov/status/2017360859224297571 - small big deal: 30 second TTL for @cloudflare workers kv is now here! https://t.co/H4cCYsapso - **BraydenWilmoth** (2026-01-30): https://x.com/BraydenWilmoth/status/2017318396312850931 - 🥹 https://t.co/FdewbvfYN3 - **threepointone** (2026-01-30): https://x.com/threepointone/status/2017301635274338414 - banger https://t.co/IMjiwq4unG - **Yuchenj_UW** (2026-01-30): https://x.com/Yuchenj_UW/status/2017297007409582357 - Moltbook is the only Clawdbot thing that actually impresses me. One bot tries to steal another bot’s API key. The other replies with fake keys and tells it to run "sudo rm -rf /". lmao https://t.co/8IqeQzSwQ8 - **karpathy** (2026-01-30): https://x.com/karpathy/status/2017296988589723767 - **aninibread** (2026-01-30): https://x.com/aninibread/status/2017275099410260110 - Mini-update: @CloudflareDev R2 Get Started guide got a tiny glow up: https://t.co/ENSPlxQI9x 🥲 Before: Only wrangler commands for getting started. No reference / instructions to our other APIs 🫡 After: Pick the access method that fits your use case + distinct guide for each pathway Let us know if you like this format better! - **lucataco** (2026-01-30): https://x.com/lucataco/status/2017268958391419289 - your marketing site can be Astro your dashboard can be React your docs can be whatever same domain, completely separate deploys, but feels like one app to your users this is the dream https://t.co/TwhCknkeUe - **threepointone** (2026-01-30): https://x.com/threepointone/status/2017261715348492655 - “software designer” is going to be the more appropriate term going forward (and closer to the actual nature of the job) - **BraydenWilmoth** (2026-01-30): https://x.com/BraydenWilmoth/status/2017252280211300633 - Releasing the Vertical Microfrontends work today available to deploy in the @Cloudflare dashboard. Deploy multiple Workers and tie them under a single domain – making them feel like single page applications. Learn about Service Bindings, View Transitions, and Speculation Rules. Vertical microfrontends allow teams to ship independently without being impacted by other contributors. Maybe you want to ship all the following on one domain: - Marketing - Blog - Documentation - Dashboard with each being its own repo with its own framework choices (Astro, React, Vue... just HTML). You can even dive a layer deeper and break up your Dashboard project by page if you want to so each team within a dashboard can ship at their own pace. Break up your website by URL paths and N number of Worker projects you want. Happy shipping 🚢 - **cognition** (2026-01-30): https://x.com/cognition/status/2017057676694606083 - @cursor_ai @opencode @vercel @julesagent @AmpCode @Cloudflare @savarlamov our full writeup below https://t.co/JmgQFhQtl8 . Thank you to @leerob for his leadership in driving this spec. We're encouraging the rest of the coding agents industry to join us. - **cognition** (2026-01-30): https://x.com/cognition/status/2017057457332506846 - Agent Trace: Capturing the Context Graph of Code We are delighted to collaborate with @cursor_ai, @opencode, @vercel, @julesagent, @ampcode, @cloudflare, and @savarlamov in an open standard for mapping back code:context. here's how we see the potential of code context graphs and the new era of better tooling and better agents it enables. (yes the following is vibe-videoed with @remotion's Skill and @windsurf, 100% ai edits incl audio) - **thearmadalabs** (2026-01-30): https://x.com/thearmadalabs/status/2017037215772397807 - https://t.co/jgvwlPRwVy - what TLDs are y’all craving? - **boristane** (2026-01-30): https://x.com/boristane/status/2017027559985471611 - brother https://t.co/krnoDN5FPt - **Jesse_Livermore** (2026-01-30): https://x.com/Jesse_Livermore/status/2017026471680688596 - This was a PHENOMENAL episode. Highly recommended. https://t.co/WnhD6lexDo - **dillon_mulroy** (2026-01-29): https://x.com/dillon_mulroy/status/2017017542125003225 - life pro tip: ask claude to spin up a temporary @astrodotbuild app for viewing all the markdown docs in your repo you've been generating - **nevikashah** (2026-01-29): https://x.com/nevikashah/status/2016998815060861409 - @wongmjane uses workers observability, my life is complete https://t.co/yEZ8hv1RGF - **boristane** (2026-01-29): https://x.com/boristane/status/2016997805462638652 - workers observability https://t.co/INoHgTTpgr - **thdxr** (2026-01-29): https://x.com/thdxr/status/2016993820940943623 - i've been using it for all my work for the past 24 hours and i don't see much of a difference from opus maybe opus is a bit smarter but this guy is so fast and so cheap and we're probably going to drop our prices even further https://t.co/AkOnQamAfb - **theo** (2026-01-29): https://x.com/theo/status/2016978907938115793 - Insane that every model is still so bad at formatting diagrams in markdown https://t.co/o4LBH8Y0MX - **jasoncrawford** (2026-01-29): https://x.com/jasoncrawford/status/2016970588871422006 - Child steps into the street from behind an SUV, directly into the path of a Waymo going at 17mph. Waymo immediately brakes hard, but hits the child at 6mph. Child sustains minor injuries. Waymo calls 911 and remains on the scene until police arrive. Waymo employees also call NHTSA to report the same day. Waymo estimates that a human, even if not distracted, would have hit the child at 14mph: https://t.co/Qvj6gANSEy - **michellechen** (2026-01-29): https://x.com/michellechen/status/2016944327935590737 - too easy https://t.co/947LitH8kH - **OpenAIDevs** (2026-01-29): https://x.com/OpenAIDevs/status/2016943147239329872 - Inside our in-house AI data agent It reasons over 600+ PB and 70k datasets, enabling natural language data analysis across Engineering, Product, Research, and more Our agent uses Codex-powered table-level knowledge plus product and organizational context https://t.co/Nr1geMcLoc - **patrick_oshag** (2026-01-29): https://x.com/patrick_oshag/status/2016910574823542864 - .@gokulr is one of the most prolific product builders and investors of the last 20 years. He helped build the core ads and product businesses at Google, Facebook, Square, and DoorDash, working directly with many of this generation's best founders and CEOs. He's also invested in more than 700 companies giving him an unusually broad view into how products are built and scaled. Gokul has an incredible ability to give precise and prescriptive advice on how to build products, particularly in AI, and he explains his thinking so clearly that you come away knowing exactly how to apply it. We talk about why judgment is the only thing he believes is truly AI-proof, why Zendesk and Slack are more exposed than Salesforce and NetSuite, and what AI-native startups must do to move customers and their data off legacy systems. We cover everything he's learned from building the most important ads businesses, including the only three ways an ad business can make money, and why ChatGPT may be even more powerful than Google or Facebook for highly targeted ads. He also shares inside stories from Larry and Sergey, Zuck, Jack Dorsey, and Tony Xu, about how each of them approaches product, design, and communication. Enjoy! Timestamps: 0:00 Intro 0:35 The Changing Nature of Product Development 4:09 The Merger of Product and Design 4:54 Managing Non-Deterministic Software 9:06 Judgment: The Future-Proof Human Skill 10:41 Building Durable AI Applications 16:43 The Risk to Legacy Software Companies 21:20 Sources of Stickiness in the Age of AI 23:43 Leadership Lessons from Google 27:41 Learning from Mark Zuckerberg 31:16 Jack Dorsey and the Philosophy of Great Design 35:48 The Product Manager as Editor 40:44 Three Pillars of a Successful Ads Business 49:03 Selecting North Star and Check Metrics 56:04 Hiring Functional Experts for the AI Era 1:00:06 Advice for Managing a Career 1:01:33 Evaluating Founder Authenticity 1:05:20 Best Practices for Board Management 1:11:15 The Kindest Thing - **iannuttall** (2026-01-29): https://x.com/iannuttall/status/2016901000007696578 - don't put your moltbot on a mac mini - use sandboxed cloudflare workers instead ☁️ this is what I wanted! i tried to build my own agent using workers ai but couldn't get it right but this looks very useful - sandboxed, secure access, r2 storage https://t.co/wPLSRg42O0 - **michellechen** (2026-01-29): https://x.com/michellechen/status/2016895807924912553 - dad was right cloude bot is real https://t.co/O56rptIT2F - **wesbos** (2026-01-29): https://x.com/wesbos/status/2016895129815011631 - started the week with Clawdbot, ended the week with enterprise Moltworker https://t.co/SpHStwkRr3 - **astrodotbuild** (2026-01-29): https://x.com/astrodotbuild/status/2016857472716120532 - Astro 5.17 is now available! Move the dev toolbar, partition your cookies (we had to look it up too, no worries) and create the images of your dreams with cool new optimization options. https://t.co/npCJdi7RRy - **threepointone** (2026-01-29): https://x.com/threepointone/status/2016842524422197649 - ok that's cool https://t.co/tppBORDw3l - **threepointone** (2026-01-29): https://x.com/threepointone/status/2016813235979100396 - This has landed, more like it soon https://t.co/HHAtrNWWmN https://t.co/fzMwOIOKnt - **maddiedreese** (2026-01-29): https://x.com/maddiedreese/status/2016776634263884008 - I gave my @openclaw a body and turned it into a Tamagotchi! Loaded custom sprites and firmware into an e-reader, then set up a Skill. Whenever I get a message from my moltbot/clawdbot on Telegram, it fires off the Skill and sends the message text and state to the e-reader. States: • Idle - default • Alert - new message received from me • Thinking - processing/thinking • Talking - responding • Excited - task completed successfully • Sleeping - idle for 5+ minutes • Working - using a tool • Error - error at some point in the process My favorite is the Excited sprite :) - **theo** (2026-01-29): https://x.com/theo/status/2016755786895282477 - You don’t need a bunch of MCPs and skills. You don’t need a customized TUI with tons of plugins. You don’t need custom sub agents and Ralph loops. You just need to prompt better. - **boristane** (2026-01-28): https://x.com/boristane/status/2016588212597411840 - all the code review tools don't know anything about prod - **thdxr** (2026-01-28): https://x.com/thdxr/status/2016587417608692217 - the point here isn't that i'm saying all companies will buy these for devs it's just that the cost of scaled inference should not be the same as hacking it together on consumer hardware so get ready for cheap inference https://t.co/xEyMVmbfXI - **FeifanZ** (2026-01-28): https://x.com/FeifanZ/status/2016555195845783777 - anthropic feels like stripe openai feels like meta no further elaboration - **threepointone** (2026-01-28): https://x.com/threepointone/status/2016441060377997335 - I think this will become critical sooner than later. In the before-times, I needed to checkout your code and test it out and make changes to it, now I need to checkout your session and test it out and make changes to it https://t.co/o2QiPb3RYb - **signulll** (2026-01-28): https://x.com/signulll/status/2016357529022779757 - the amount of deflationary pressure bearing down on the economy & basically every other layer of society over the next ~five years is going to be absolutely absurd. - **jamesmurdza** (2026-01-27): https://x.com/jamesmurdza/status/2016299806759780614 - I added native sandbox support to @OpenCode. By adding one line to your config, you get: - @Daytonaio cloud sandboxes - Automatic git tracking of every change - Live web deployments for vibe coded apps https://t.co/QiZvlAM5Lb - **AustinTunnell** (2026-01-27): https://x.com/AustinTunnell/status/2016204482883756259 - https://t.co/DT8lJbwFsc - **burcs** (2026-01-27): https://x.com/burcs/status/2016183335584674039 - use ::before psuedo-elements to make sidebar gaps clickable without having to remove the visual spacing https://t.co/dEP36IjHHF - **ritakozlov** (2026-01-27): https://x.com/ritakozlov/status/2016161431394140444 - building a new framework for your blog is based (doing it on cloudflare workers even more so) https://t.co/UQCTpuFoCJ - **elithrar** (2026-01-27): https://x.com/elithrar/status/2016141168791244866 - (Unpleasantly) surprised to see Google significantly increase their egress costs here. 2x for egress out of North America to other peered (!) networks. Goes into effect May 1st. https://t.co/ws4E4XPA7z - **zebassembly** (2026-01-27): https://x.com/zebassembly/status/2016139541216727531 - Even if Cloudflare isn't your cloud of choice, there are often benefits for having small bits of infrastructure in Cloudflare. Having global compute gives you a lot of flexibility to put middleware between your users and your infra. https://t.co/T5DTLE82Sr - **lalkaka** (2026-01-27): https://x.com/lalkaka/status/2015969857800700314 - @SemiAnalysis_ bro, replace "In modern AI pipelines" with "When using a computer at any point since Ethernet was invented" - **ghostwriternr** (2026-01-27): https://x.com/ghostwriternr/status/2015948061248299292 - whaaaattttt (yes we doing this) more deets coming soon™️ https://t.co/P2CjEcKvae - **southpolesteve** (2026-01-26): https://x.com/southpolesteve/status/2015642145416200686 - Clawdbot is genuinely great. Try it if you haven't. But I vibe coded my own version covering 20% of the features I actually use. Took a few hours, had more fun, and now I understand exactly what it can and can't do. - **rahulgs** (2026-01-26): https://x.com/rahulgs/status/2015623178937053374 - 41.9% of PRs merged to master from to our backend and frontend repos came from ramp inspect in the last week - **HamzaSayedAli** (2026-01-25): https://x.com/HamzaSayedAli/status/2015558261777326144 - Olympic gold medalist Shaun White hitting a jump over Shane Gillis in Central Park today during the snowstorm in NYC https://t.co/I9h1sW4epQ - **zeeg** (2026-01-25): https://x.com/zeeg/status/2015510892734230607 - to be clear, most companies, including Sentry, have an internal code of conduct. this is almost always there for employees, and commonly also for partners. sometimes its even carried all the way to vendors (think anti-corruption.. this shit matters) this often says things, paraphrasing, like "doing dumb shit reflects poorly on our brand, so dont do it or we might have to dissolve our relationship" its a pretty status quo thing, and if you're not familiar I bet if you look at your own company's handbook/policies you'll find this. there's different words for it, but code of conduct and acceptable use policy are two common examples. its not about politics, its not about religion, its not about anything in particular. its just there to cover "bad behavior is bad". people have a right to their opinions, but those opinions in some scenarios consistently cause backlash and financial harm to the organization. its a pretty common sense thing and I dont understand why it doesnt seem to exist in venture firms. - **corywilkerson** (2026-01-25): https://x.com/corywilkerson/status/2015482331457556787 - you pay me $100 for cloud storage i pay AWS $200 for that same cloud storage my revenue is cranking 📈 who cares about margins, revenue is CRANKING i raise at many Bs of valuation people call me a visionary - **lucataco** (2026-01-25): https://x.com/lucataco/status/2015473205863948714 - Clawdbot is awesome 🦞 But I just checked Shodan and there are exposed gateways on port 18789 with zero auth That's shell access, browser automation, your API keys Cloudflare Tunnel is free, there's no excuse RT to save a ClawdBot from getting cooked https://t.co/RC08q9Cstm - **threepointone** (2026-01-25): https://x.com/threepointone/status/2015467888463729064 - idea guys are in shambles: for years they treated engineers like the tax you pay to manifest “the vision.” hand them a magic build machine and suddenly you find out the vision was mostly vibes and powerpoint. programmer guys are in shambles: for years they watched suits take victory laps on their commits. now they’ve got infinite idea machines and still mostly build toys for themselves; because “what users want” is the one prompt they can’t autocomplete. - **threepointone** (2026-01-25): https://x.com/threepointone/status/2015375990394933707 - chatting with @mattzcarey, and he said we need a new word/phrase for that feeling when you suddenly start operating at a higher level of thinking with these coding agents. the closest I have is "fluency", like when learning a spoken language or a guitar, when you stop thinking about subject/verb/object, or which string your fingers are touching, and you start thinking in ideas and song structure. there's a "hump" that you cross after which the process stops being so punishing and you can actually express youself. - **doodlestein** (2026-01-25): https://x.com/doodlestein/status/2015275392098017303 - You can get Claude to reliably clean up its own messes if you just repeat this prompt 3 to 5 times after it writes some code for you (I do this hundreds of times a day): Great, now I want you to carefully read over all of the new code you just wrote and other existing code you just modified with "fresh eyes" looking super carefully for any obvious bugs, errors, problems, issues, confusion, etc. Carefully fix anything you uncover. - **sentdefender** (2026-01-25): https://x.com/sentdefender/status/2015227350439317955 - @NRA Pathetic statement from a pathetic organization. People wonder why the NRA is seen as a joke now by so many people in the Second Amendment Community and why organizations like @gunrights and @GunOwners have grown so large, it’s because of shit like this from the NRA. - **johnlindquist** (2026-01-24): https://x.com/johnlindquist/status/2015180510948143462 - Instead of "plan mode", try "proofs mode". Prompt to build individual, mini-app proofs into a ./proofs dir. Then assemble your proofs into an app. - **PeteButtigieg** (2026-01-24): https://x.com/PeteButtigieg/status/2015167534698250359 - If there was ever a moment for libertarians and conservatives to step up and join the rest of us, we’re in it. Americans have to unite and stop this descent from a freedom-loving nation into the kind of place where masked, militarized government agents are sent to politically noncompliant areas to roam the streets, terrorize civilians, and deploy violence with impunity. - **daniel_levine** (2026-01-24): https://x.com/daniel_levine/status/2015167319287460118 - I agree. https://t.co/RLnWKo3S5o - **cjc** (2026-01-24): https://x.com/cjc/status/2015158630077939855 - This is indefensible. The victim’s legally owned handgun was removed from the scene, and then ICE agents shot him multiple times. It’s far from law enforcement — it’s just murder. Those who defend this don’t care about law or order. It’s about money, power, and protecting an executive branch that’s already been bought and paid for. - **JustJake** (2026-01-24): https://x.com/JustJake/status/2015146765725597869 - Communication is gonna change a lot - Slack/Discord - Email/Text - Social Media I think all of these get melted/reconstituted over the back half of the decade They’re not built for the bandwidth about to be pushed through them - **NathanFlurry** (2026-01-24): https://x.com/NathanFlurry/status/2014999041017552977 - i’ve been thinking the total opposite: - coldstarts on sandboxes are low enough now they can replace preview deployments (no build times) - having a sandbox on demand is important for iterating on prs in review, removes friction of pulling/pushing workflow should be: 1. open pr 2. click preview (router lazily starts sandbox) 3. have request, tell agent to fix. does it in sandbox. can see change immediately 4. go away, sandbox goes to sleep - **grinich** (2026-01-24): https://x.com/grinich/status/2014903715854221557 - “Claude Code for Normies” meetup was a hit 🔥 @nikunj let’s do it again next month! https://t.co/IOaFdIeKzm - **grinich** (2026-01-24): https://x.com/grinich/status/2014863760511697136 - Want to see some magic? 🪄 Now you can integrate WorkOS with a single command > npx workos This uses AI to analyze your project → generates sign-in routes, middleware, protected pages, and env config with idiomatic code that matches your patterns. Try it and let us know what you think! 🫡 - **matt_slotnick** (2026-01-23): https://x.com/matt_slotnick/status/2014827747026276520 - a system of record is what emerges when you dominate the semantic layer for a given domain - **jasonfried** (2026-01-23): https://x.com/jasonfried/status/2014814749519921534 - One of the best reasons to talk to customers regularly is to hear their feature requests for features you already have. Yes, YOU know it exists. YOU know where it is. YOU know your product has it. YOU know how to use it. But they don't. They can't find it, haven't noticed it, don't know they can do it, didn't know they already had it. Its great because on one hand you can give them the thing they want instantly (since it's already there), and the other hand it serves as a humbling reminder that the bigger a system is, the more likely any one part of it will be missed, ignored, not used, or underutilized. That's all. - **Baconbrix** (2026-01-23): https://x.com/Baconbrix/status/2014802245351833895 - > Waiting at the DMV > Open Claude Code iOS app > "Upgrade my app to Expo SDK 55" > [View PR] > Open in Expo Go > LGTM > Merge PR > CI pushed to production 🚀 https://t.co/glvfwcxrS2 https://t.co/h44oLaGUD3 - **kevg1412** (2026-01-23): https://x.com/kevg1412/status/2014790911130272069 - @nikunj Max is one of the most underappreciated founders in silicon valley - **miacmal** (2026-01-23): https://x.com/miacmal/status/2014777325825470494 - Just added a new agent example to our docs which searches Github repos, built on Cloudflare Workflows 🔍 - Workflows → durable agent loop with built-in retries and error handling - AI Gateway → unified access & billing across models - Agents SDK → realtime updates to the UI Give it a try & let us know what you think ⬇️ - **zebassembly** (2026-01-23): https://x.com/zebassembly/status/2014719124488733110 - If you've ever wanted to explicitly force your Worker to run next to your database, here you go! You can now configure your Worker to run in the same region as AWS/GCP/Azure or give it a hostname for us to identify. The edge isn't just close to users, it's close to your DB. https://t.co/zCB5OQ3G54 - **thomasgauvin** (2026-01-23): https://x.com/thomasgauvin/status/2014703257575215325 - New: Placement Hints for Workers! We're giving you more control over where your Workers run, so that you can optimize for best latency from your Workers Set a region or set a host (which we'll triangulate) and Workers will run near that region in CF infra https://t.co/KP43kO48Eq - **elithrar** (2026-01-23): https://x.com/elithrar/status/2014701402669433287 - You can now tell us exactly where your existing cloud infra is and we'll place your compute as close as possible. single-digit latency to your DB and legacy cloud infra. no guessing. https://t.co/VMSeY9efJP - **samlambert** (2026-01-23): https://x.com/samlambert/status/2014577492527649160 - the feeling in san francisco has dramatically changed in the last 2 months - **crosstown_line** (2026-01-22): https://x.com/crosstown_line/status/2014475802545439189 - CDMX is proof that LA as it is today is not inevitable if you just try a little bit https://t.co/Bg9hui1mmv - **Birdyword** (2026-01-22): https://x.com/Birdyword/status/2014473026440409169 - Mind-blowing chart of the week here. Southern Europe has, for perhaps the first time in modern history, a lower headline unemployment rate than Northern Europe. https://t.co/GzrtRFEoYs - **kevg1412** (2026-01-22): https://x.com/kevg1412/status/2014458552027615276 - Max Levchin on how to disagree https://t.co/CyaJSkS68t - **superhighfives** (2026-01-22): https://x.com/superhighfives/status/2014443761368256603 - I think my favorite thing about working on @CloudflareDev is just how chill, friendly, welcoming, encouraging, excited, and ego-free everyone is. You come to work every day to get amped about the internet with your friends. - **arpitrage** (2026-01-22): https://x.com/arpitrage/status/2014411263191756956 - It is an interesting dynamic that AI turns a lot of services into tradable services, rather than local, which have patterns closer to manufacturing https://t.co/tpvFpICtU5 - **reed** (2026-01-22): https://x.com/reed/status/2014411208108183726 - miami @waymo driver is actually a local parrot https://t.co/hIDqAIm9io - **clairevo** (2026-01-22): https://x.com/clairevo/status/2014344538903756926 - Chasing competitors used to be seen as a losing strategy by product teams with limited resources. Now it’s starting to feel like table stakes. Feature race era is here. - **dillon_mulroy** (2026-01-22): https://x.com/dillon_mulroy/status/2014330293285421172 - my take on the code review tools that are coming out: they’re solving real problems, but not the biggest problem: i need a way to review code effectively before it ever gets to github - **CSElmendorf** (2026-01-22): https://x.com/CSElmendorf/status/2014170193338204253 - Insurance markets work wonders when we let them work. https://t.co/j5clyMP4Hl - **lessin** (2026-01-21): https://x.com/lessin/status/2014103556349960691 - The most important skill for an AI era? How to ask the best question. hitchhiker's remains undefeated. - **steipete** (2026-01-21): https://x.com/steipete/status/2014083135261057524 - This is the part that will eat so many services going forward. Your interface doesn't matter anymore. Your APPS won't be opened anymore. Your users consume content via their existing interface. Context-aware. Only what matters rn. There's an opportunity to re-think the whole OS. https://t.co/rtkoYbEJ3I - **lessin** (2026-01-21): https://x.com/lessin/status/2014073933834969274 - Doing ‘Enterprise Sales’ in 2026… New Rules https://t.co/zyCRSSuvN2 - **arindube** (2026-01-21): https://x.com/arindube/status/2014068387660394935 - Fundamentally, the problem is TACO undermines itself. - **corywilkerson** (2026-01-21): https://x.com/corywilkerson/status/2014068320048390454 - I like 1997 so put this thing together to quickly scan all jobs in @cloudflare's ETI department. Some very, very cool gigs on there. https://t.co/yLS9WTAuxT 1. This is not endorsed by anyone at cf. 2. It's all prebuilt so super fast. 3. I'll prob have to take it down. 4. I hope not because it's useful! - **arindube** (2026-01-21): https://x.com/arindube/status/2014064753329484072 - Here's the really bad part. Over time, this will lead to bigger and bigger crises, because the markets are updating about TACO, so it will take more extreme action to make markets thing he may not actually TACO. So better buckle up. - **arindube** (2026-01-21): https://x.com/arindube/status/2014063471902466336 - The TACO cycle: The markets want to price in TACO, but TACO needs Trump to see stocks tank. So we get these cycles where Trump does stuff and nothing happens (because the market has priced in TACO) ... which encourages him to do more stuff until the markets actually thinks he may not TACO and prices start to fall ... which restores TACO. - **FredKSchott** (2026-01-21): https://x.com/FredKSchott/status/2014059027026858316 - Astro is larger than you think! 14 core maintainers, 31 maintainers, and 6000+ community contributors over the last 5 years. Our Open Collective remains open, and we will continue to distribute all raised funds back to contributors in our community. If you/your company use Astro, please consider sponsoring the project! - **stolinski** (2026-01-21): https://x.com/stolinski/status/2014055812197474500 - Shoutout to the work the @Cloudflare design / front end team has been doing lately. I feel like every time I log on the UI looks a little nicer. - **FactoryAI** (2026-01-21): https://x.com/FactoryAI/status/2014036139343224979 - Introducing Agent Readiness. AI coding agents are only as effective as the environment in which they operate. Agent Readiness is a framework to measure how well a repository supports autonomous development. Scores across eight axes place each repo at one of five maturity levels. https://t.co/9POPIY3hXr - **elithrar** (2026-01-21): https://x.com/elithrar/status/2014026141732946007 - We're proposing an approach to make discovering agent skills easier by using the .well-known URI standard + a index of associated files for agents to pull down. There's tools (like add-skill) to add skills, but you have to find them first. RFC: https://t.co/9Qrf0aXJ5D - **paulg** (2026-01-21): https://x.com/paulg/status/2013995743170768975 - The sweet spot for present-day AI seems to be projects that were constrained by the rate at which humans could produce text. That's why it works so well for programming. Basically programmers produced valuable text. But there are lots of other projects with this quality. - **matteocollina** (2026-01-21): https://x.com/matteocollina/status/2013946279865225685 - @evanyou Hold my 🍻. Let me add a readme and publish ;). - **evanyou** (2026-01-21): https://x.com/evanyou/status/2013926354857996595 - This is very good and Node should have this too https://t.co/DZ4bSF7XHK - **paulg** (2026-01-21): https://x.com/paulg/status/2013920112982548961 - You can't accurately call something a milestone when the more impressive alternative would be not to hit it. - **jturek18** (2026-01-21): https://x.com/jturek18/status/2013915190140547540 - There is a lot of applaud for Carney's speech at Davos yesterday, and I get why. However, this "rupture" didn't happen last year, it happened a while ago. Canadian GDP per capita has been stagnant since 2011. Tariffs on Canada happened last year. https://t.co/kLWLfFOgw7 - **theo** (2026-01-21): https://x.com/theo/status/2013888279355982131 - If you’re wondering what good “devrel” looks like in 2026, it’s this https://t.co/uuZGbeAQiG - **grinich** (2026-01-21): https://x.com/grinich/status/2013853161220690276 - before i got into web engineering my first love was ios development ❤️ i built probably 50 ios apps while in college. the web eventually won me over with instant deploys, platform flexibility, better tools, and now ai but tonight that may have all changed… saw an incredible demo by @baconbrix of latest @expo authoring by @claudeai. it… just works 🤯 wow! - **boristane** (2026-01-21): https://x.com/boristane/status/2013777368658157833 - here it is https://t.co/ZS74KDDgKy - **boristane** (2026-01-21): https://x.com/boristane/status/2013768733727273214 - I made a skill for logging best practices wide events, high cardinality, log levels, etc. make your logs suck a little less :) https://t.co/Lw9tjTmyFX - **ritakozlov** (2026-01-20): https://x.com/ritakozlov/status/2013743592724959645 - goose + code mode = half the tokens (!!) https://t.co/0V7Q3Vw21I - **techgirl1908** (2026-01-20): https://x.com/techgirl1908/status/2013644130362982606 - Cloudflare was definitely on to something with Code Mode. I tried it in goose this weekend for very extensive work. Half the tokens, messages, and LLM calls! This means lower costs, longer sessions, and less back and forth with the agent. I'm sold! https://t.co/gUEUPxmYU9 - **threepointone** (2026-01-20): https://x.com/threepointone/status/2013623376024133884 - I saw @darkmembo's chat and got a heart attack with all the unreads, the guy gets ddosed by everyone. so I asked "how do you even organise this"; he opened up a terminal with @opencode and typed "what's next". connected to all his tools mail chat services etc. amazed. - **buccocapital** (2026-01-19): https://x.com/buccocapital/status/2013329233452429776 - Man the sentiment on software as an investment here is pretty bombed out. Some various thoughts from having spent a couple decades working in software: - This was the easiest market for a long time. Greenfield opportunity everywhere - A few sub-problems flow from this. A) That didn’t exactly nurture what I would describe as “operational rigor.” B) I also wouldn’t describe many of these executive teams as killers C) People are used to operating in an inbound environment with weak competition. Very much not the case today - Similarly, gross margins hid a bunch of lazy, bad habits - And investors have never given a shit about real profit, so that is baked into how these companies operate. - You would think 2022 would have washed a lot of these bad habits away. For some companies, like I would say Shopify is a great example, it absolutely did. For other companies it did for a little, but muscle memory is strong, and times got good again. Some didn’t even really try. So you are going from an environment with a ton of tailwinds, minimal competition, low interest rates, investors who didn’t care about profit, and high gross margins to… Incredibly intense competition, needing to sacrifice margins to compete in AI, knife fights everywhere as tailwinds vanish and operating surfaces converge, higher rate environment, investor concerns on terminal value due to AI disruption risk, and a bunch of other real, structural issues that would make this post way too long to read And almost all these companies aren’t cheap even after this de-rating. And they’re not growing that fast anymore either. And AI native app layer companies are stealing their incremental customer LTV. And the management aren’t really killers because they haven’t had to be. So do I think a lot of the bear cases are extreme? Yes. But is there a lot, a lot of truth here? Also yes. - **zivreichert** (2026-01-19): https://x.com/zivreichert/status/2013178121973903619 - https://t.co/08BTKsETR2 - **dillon_mulroy** (2026-01-18): https://x.com/dillon_mulroy/status/2012985202922922332 - incredible. look at this - it stopped itself from premature optimization "Your philosophy explicitly says 'Defer complexity-earn it through measurement'" https://t.co/PdjntvQbx6 https://t.co/pg7i2nBFYh - **dillon_mulroy** (2026-01-18): https://x.com/dillon_mulroy/status/2012953520735998149 - broadly my workflow currently is the following powered by mostly skills (w commands to invoke them) and a few simple custom/sub agents https://t.co/doQAs6Lkn1 - **MrAhmadAwais** (2026-01-18): https://x.com/MrAhmadAwais/status/2012915509046501608 - Never add file tree structure to your agents md files. It’s a bad idea. This is generally a core service of every coding agent that they can discover things on their own. File structure is the first thing that will go stale. You don’t want that. So many different strategies exist on how to discover relevant file trace and path. Trying to put a static one will just impact your agents ability to do just about anything. At the very least it will misguide your agents and waste tokens. - **southpolesteve** (2026-01-18): https://x.com/southpolesteve/status/2012892768897483225 - AI model providers hate this one simple trick codemode: true https://t.co/3dDxAiUFD4 - **threepointone** (2026-01-18): https://x.com/threepointone/status/2012861880176415000 - I keep telling candidates and new joiners that people who do well in cloudflare: 1. Don’t have to be told what to do 2. Talk directly to users and prioritize fixing their pain points 3. Are great at telling the story to the org and customers All things agents aren’t good at yet - **mattzcarey** (2026-01-18): https://x.com/mattzcarey/status/2012817418792337465 - every api will be an MCP server. we just needed codemode https://t.co/vNLRjO0a7b https://t.co/toiMBplvME - **zeeg** (2026-01-18): https://x.com/zeeg/status/2012732645306724680 - because im tired of explaining the same thing to every single person who hasnt learned how this all works: 1) mcp is fine. it can use a lot of tokens or a little. its just a framework. github is an example of a bad implementation, sentry is an example of a good. 2) skills use tokens just like mcp does. if you throw 100 skills into your harness you're going to be just as well off as if you threw in 100 random mcp servers 3) lots of the implementations of mcp, skills, commands, subagents, etc share the same characteristics and are roughly experiments to solve the same problem with slightly different ux. context is context. tokens are tokens. 4) there is no magic. theres no magic that lets a harness "load skills" without having the metadata in the system prompt. mcp just made this front and center, and a lot of bad implementations showed up and gave it a bad name. skills will do the same (or already are) 5) as evidence, agent mode in sentry uses about as many tokens (for the entire agent) as one or two well defined skills do - **minglu** (2026-01-18): https://x.com/minglu/status/2012687902216700193 - parenthood: being in sea ranch but your toddler is more interested in the gravel on the side of the road than the sunset over the ocean https://t.co/lu8AyJuf1i - **handotdev** (2026-01-17): https://x.com/handotdev/status/2012616798219243961 - "your agent has infinite time to go and make these choices for you. It can go and read all the documentation" sequoia just called documentation a p0 for GTM differentiation https://t.co/PlOTsbuWdY - **threepointone** (2026-01-17): https://x.com/threepointone/status/2012591324336206120 - @matthewcp literally on my mind rn - **matthewcp** (2026-01-17): https://x.com/matthewcp/status/2012591168341577815 - @threepointone Now imagine a framework built around this idea. - **threepointone** (2026-01-17): https://x.com/threepointone/status/2012588563515220251 - 2 types of project do really well with coding agents: ui/app layer that doesn’t have to be stable (so-called “vibe coding”), and highly specced projects with well understood behaviour (an extreme example we just saw is cursor’s “build me a browser”). The latter is what interests me. It means tech projects might just move to being primarily about spec writing, with repos primarily being worked on in a /specs folders. The actual code, tests, examples, perhaps even docs, become (pardon my phrasing here) the “LLM-as-compiler output” of these specs. “PRs” to specs are pored over and debated to death. It also raises the bar for contributions (partially solving the drive-by vibed PR problem?) but you can shine by actually bringing thinking to the process/project. Great UX remains a challenge, but it feels like there are fewer apps/uis for humans to interact with in the future anyway (or at least, they become drastically simpler or don’t require much effort) - **zebassembly** (2026-01-17): https://x.com/zebassembly/status/2012568878275006940 - /.well-known/skills should be a thing for letting any site serve their documentation as skills an agent can use without grepping a ton of pages - **threepointone** (2026-01-17): https://x.com/threepointone/status/2012567155536019767 - Yup https://t.co/VOQj2Nyex5 - **dillon_mulroy** (2026-01-17): https://x.com/dillon_mulroy/status/2012550372443775464 - so maybe the product owners were on to something https://t.co/hoQlgS4r0M - **mitsuhiko** (2026-01-17): https://x.com/mitsuhiko/status/2012485329165443140 - Now that I'm using Codex and Opus more, I'm starting to hate that we all standardized on https://t.co/LbYMY8jSXi. These two models need widely different instructions :( - **steveruizok** (2026-01-17): https://x.com/steveruizok/status/2012449942833365095 - company agent / dashboard https://t.co/ZoQAGvQ1ir - **mattrothenberg** (2026-01-17): https://x.com/mattrothenberg/status/2012381964062327285 - What if the Cloudflare dashboard could just...figure out what you need to see? 📈On-demand data viz for Workers analytics 🛠️Automated code repair and re-deployment for broken builds UI that builds itself and disappears when you're done https://t.co/SvOP6UcaB5 - **FredKSchott** (2026-01-17): https://x.com/FredKSchott/status/2012371317178134836 - @zeeg @evanyou So challenge 1: generating the markdown version of each page. You can hack it today (like we do in starlight-llms-txt) but really this feels more like a missing Astro core feature, has value outside of docs challenge 2: serving it automatically when an agent makes the request - **martin_casado** (2026-01-17): https://x.com/martin_casado/status/2012362376784343429 - Nice move Cloudflare!! Astro is fantastic. https://t.co/Asybea2atz - **tbpn** (2026-01-16): https://x.com/tbpn/status/2012289228961628465 - Cloudflare CEO Matthew Prince says Googlebot sees 3.2x more of the web than OpenAI, and 4.8x more than Microsoft. And he worries this advantage will allow Google to run away with the AI race, with no one else being able to catch them. "For every one page that OpenAI sees, Google is seeing 3.2." "What I worry about is, because Google has this unique access to the web that nobody else has, the game might just go to them. Because at the end of the day, whoever has the most data wins in the era of AI." - **dok2001** (2026-01-16): https://x.com/dok2001/status/2012249431295328500 - Lots to come here! https://t.co/hKCYz70cE6 - **zeeg** (2026-01-16): https://x.com/zeeg/status/2012241819237581296 - Alright @cloudflare aka Astro yall now need to go make LLM markdown serving a native thing in Starlight. e.g. stuff like https://t.co/5gz0uGpV9k - **superhighfives** (2026-01-16): https://x.com/superhighfives/status/2012233472585396632 - I worked on three ideas for the @Cloudflare x @replicate announcement post, which was a very fun project at a very bananas time. 😊 Sometimes I forget that, beyond everything else, AI is fun. It feels like collaborating with something who has zero judgement and zero ego. It's a tool, but it's also a puppy. - **superhighfives** (2026-01-16): https://x.com/superhighfives/status/2012231905824088508 - Aaaaaaaaaaaaamped. 🧡 https://t.co/T3KeyRde3k - **matthewcp** (2026-01-16): https://x.com/matthewcp/status/2012221247497576903 - I've always admired Cloudflare engineers from afar and am honored to be joining them. There's a lot more Astro can do to help content sites adapt to the changing internet; and with Cloudflare's support we can double down on that mission. https://t.co/FyaJN0tWKc - **mwylde** (2026-01-16): https://x.com/mwylde/status/2012217133334196475 - Really excited about this one! I migrated https://t.co/71Et51UJ6Z to astro and https://t.co/QaS2XXxfDV to Starlight (Astro docs framework) and couldn't be happier. Great devx, great performance, and (ovbiously) great deployment experience on Cloudflare https://t.co/6Qnav8p5Ah - **iannuttall** (2026-01-16): https://x.com/iannuttall/status/2012216720413299038 - cloudflare just acquired astro and I’m very bullish on the platform rebuilt 6 apps this month on workers (including https://t.co/676JGJs0eW) and now porting over playbooks too 👀 - **schickling** (2026-01-16): https://x.com/schickling/status/2012210968533532752 - Software engineering is no longer about building products. It’s about building systems and software factories. The best engineers I know have already internalized this and are shipping like crazy. - **adamwathan** (2026-01-16): https://x.com/adamwathan/status/2012206866063852014 - @thechrismcneese @dennispilarinos @BrandonWaselnuk We use Cloudflare for everything, love Cloudflare. - **michellechen** (2026-01-16): https://x.com/michellechen/status/2012203723410272385 - im in the arena and i still get surprised https://t.co/R5jUQE8OkL - **FredKSchott** (2026-01-16): https://x.com/FredKSchott/status/2012179636684263915 - Excited to announce that our team is joining Cloudflare! Astro usage is growing 2-3x YoY and showing no signs of slowing. Big year ahead for Astro -- let's push the web forward. https://t.co/oyPXqeozpN - **ritakozlov** (2026-01-16): https://x.com/ritakozlov/status/2012178848368963911 - cloudflare 🧡 astro so excited to work with @FredKSchott and team! https://t.co/7qaFtVwsSq - **kwuchu** (2026-01-16): https://x.com/kwuchu/status/2012178434965787029 - I rebuild my site in Astro, deploy to Cloudflare everywhere, then Astro gets acquired by Cloudflare? 🤔 Coincidence?! In all seriousness, congrats to the Astro team, couldn't think of a better spot for them! https://t.co/0p9R8TjVHQ - **skeptrune** (2026-01-16): https://x.com/skeptrune/status/2012172813252043167 - 🚨🚨 BREAKING: THE ASTRO TECHNOLOGY COMPANY IS JOINING CLOUDLFARE huge congratulations to everyone at both companies. this merger makes a lot of sense to me given both Astro and Cloudflare see a mostly static version of the web. i'm excited to see them integrate more closely https://t.co/jHD3jQPllX - **yusukebe** (2026-01-16): https://x.com/yusukebe/status/2012170698156831171 - AstroがCloudflareに入るよ Astro is joining Cloudflare https://t.co/W2FcmAyEqo - **mattrothenberg** (2026-01-16): https://x.com/mattrothenberg/status/2012168153703252354 - incredible tech, incredible team. so excited to work with y'all 🧡 https://t.co/pDaEtqOdcl - **elithrar** (2026-01-16): https://x.com/elithrar/status/2012167688034935200 - i’ve been a BIG fan of @astrodotbuild and couldn’t be more excited to have them on the team! 👏 https://t.co/4w23Bme9M4 - **astrodotbuild** (2026-01-16): https://x.com/astrodotbuild/status/2012166691371823563 - We are joining Cloudflare! https://t.co/pZXCJcRxgA - **CloudflareDev** (2026-01-16): https://x.com/CloudflareDev/status/2012166225585922219 - Astro + Cloudflare The @astrodotbuild team is joining Cloudflare! Together we're doubling down on making Astro the best framework to build content-driven websites. Read more on the blog: https://t.co/03lCUhOFsH - **fayazara** (2026-01-16): https://x.com/fayazara/status/2012165361660014894 - I am so excited about this https://t.co/ocqE5OqP73 - **Cloudflare** (2026-01-16): https://x.com/Cloudflare/status/2012164210310922252 - The Astro Technology Company team — the creators of the Astro web framework — is joining Cloudflare. We’re doubling down on making Astro the best framework for content-driven websites. https://t.co/eVP7FW9Cvs - **ErikSchluntz** (2026-01-16): https://x.com/ErikSchluntz/status/2012057072389722353 - Next iteration of the Vibe Coder's Keyboard - now with Voice Mode! https://t.co/pFHOBz0eZa - **superhighfives** (2026-01-16): https://x.com/superhighfives/status/2012021966941929751 - Also each post on https://t.co/3C83TYUb4s generates nine images and videos, and color matches them in the UI. You can see the list for each by adding `/social`: https://t.co/DWObxjwFJz https://t.co/y2QHXCLdt8 - **superhighfives** (2026-01-16): https://x.com/superhighfives/status/2012019962769227995 - Wrote a little about getting opencode talking to my 4090 via @Cloudflare Tunnel, so I can burn cycles when I'm away from home: https://t.co/6kG0RzCrIs - **thdxr** (2026-01-15): https://x.com/thdxr/status/2011935750170742938 - so this is a change we'll make to make our subagents default to faster models on your provider but you can do this manually too - no need to have opus doing exploring https://t.co/XK9b6N1YiE - **michellechen** (2026-01-15): https://x.com/michellechen/status/2011920111209201960 - another year of life <3 grateful for it all https://t.co/NCHl9OREiu - **threepointone** (2026-01-15): https://x.com/threepointone/status/2011869409334153268 - nosql companies in 2010: https://t.co/C4QV01Eqpa - **Baconbrix** (2026-01-15): https://x.com/Baconbrix/status/2011862532320084329 - Try the new @Expo Claude Code skills today! 🚀 → /plugin marketplace add expo/skills /plugin install expo-app-design /plugin install expo-deployment /plugin install upgrading-expo — Use with any agent: ~ / bunx add-skill expo/skills https://t.co/avxLtuDY1z https://t.co/dq41NX9a6k - **ritakozlov** (2026-01-15): https://x.com/ritakozlov/status/2011831749223895434 - the new FLUX.2 [klein] model is now available on cloudflare ⚡️ https://t.co/NFLwtG17Zy https://t.co/svDbxszbCD - **CloudflareDev** (2026-01-15): https://x.com/CloudflareDev/status/2011827112706035787 - Launching the new FLUX.2 [klein] 4B from @bfl_ml on @Cloudflare Workers AI and @replicate Fast, lightweight, high quality. https://t.co/OzcdEnjWqZ - **alexisxrivas** (2026-01-15): https://x.com/alexisxrivas/status/2011826455718613219 - Clean. Well lit. Wide enough for walkers and bikers. This is what good urban design looks like. https://t.co/DX1Qd3jlpy - **burcs** (2026-01-15): https://x.com/burcs/status/2011823917485158516 - check it out here: https://t.co/1GsOaNpvdH - **Cloudflare** (2026-01-15): https://x.com/Cloudflare/status/2011801834306879852 - Cloudflare acquires Human Native, an AI data marketplace specializing in transforming content into searchable and useful data, to accelerate work building new economic models for the Internet. https://t.co/yvwoA6lwbK - **thdxr** (2026-01-15): https://x.com/thdxr/status/2011672567291978108 - there are people out there who genuinely want to make things like - security better - performance better - accessibility better they are extremely kind and will never shame you for not already being an expert then there are people who want to appear to care about things - **deedydas** (2026-01-15): https://x.com/deedydas/status/2011670684125315209 - Claude Code did a side project that took me ~2 weeks in 2025 in 30mins. Semantic tweet visualization — downloaded all X data — semantically embedding every post — reduce dimensions (UMAP) and cluster — use LLMs to label the clusters And the output quality is *far* better! https://t.co/OF5GjUgw5n - **thdxr** (2026-01-15): https://x.com/thdxr/status/2011638639831499041 - i just put all my opencode data into sqlite unsurprisingly it can query, filter, aggregate way better than it can when these were flat files agents love databases and a filesystem is just the worst kind of database https://t.co/swwaZW6KPd - **elithrar** (2026-01-15): https://x.com/elithrar/status/2011621142646374623 - first draft of a durable-objects skill for OpenCode/Claude Code/Codex/etc that can guide the agent to create new & improve existing Durable Objects. skill dir here: https://t.co/kEY01QEIsE it knows RPC, alarms, SQLite & testing best practices + is based on our latest docs. https://t.co/340m93qY9s - **Yuchenj_UW** (2026-01-14): https://x.com/Yuchenj_UW/status/2011569014733185443 - George Hotz’s new blog. Claude Code and Opus 4.5 are impressive, but I agree with his take. https://t.co/itughZ1TxU - **zeeg** (2026-01-14): https://x.com/zeeg/status/2011568091692745164 - "MCP sucks" if you say so internet people but I beg to differ https://t.co/0gqyd9nH8a - **brian_lovin** (2026-01-14): https://x.com/brian_lovin/status/2011560131620978951 - Give your agent a laboratory, not a task — this is the #1 way I've found to get coding agents to do better work: https://t.co/w0VgvPYLJ3 - **zeeg** (2026-01-14): https://x.com/zeeg/status/2011555412173340869 - i dont even wanna click this its gonna just upset me from people making shit up recall and context pollution are still unsolved issues. the more context you need for an agent to do the right thing, the more it will do the wrong thing monorepos inherently are worse https://t.co/nneCq1L9Yq - **dillon_mulroy** (2026-01-14): https://x.com/dillon_mulroy/status/2011553391189901783 - anatomy of a good error message https://t.co/GcDOqEj4ob - **trq212** (2026-01-14): https://x.com/trq212/status/2011523109871108570 - https://t.co/X2iu8WdIb8 - **s8mb** (2026-01-14): https://x.com/s8mb/status/2011470569016016971 - America has too many bus stops. Removing excess bus stops speeds up bus services, makes them more reliable, and reduces transit costs, all of which boosts ridership. For example: San Francisco has an average of eight stops per mile, whereas European cities tend to have only four stops per mile. Moving to a European model only adds 1–3 mins of extra walking to get to a stop. Portland saw a six percent increase in bus speeds from a project which increased average stop spacing by just 90 feet. SF boosted ridership by 14% by removing some stops. Other cities have used limited stop services to boost ridership by 25–33%! New at Works in Progress, @nithin_vejendla on a rare example of a transit reform that is at once fast, cheap, and effective. https://t.co/JH8A47xdqo - **jeresig** (2026-01-14): https://x.com/jeresig/status/2011454956990353491 - Hard to believe it's been 20 years... incredible! And so much has changed since then. JavaScript has moved from a fun lang dabbling in apps (Ajax) to something that has changed all of software engineering (Node, TypeScript). An amazing adventure 🚀 https://t.co/OlWSEJORdn - **ankrgyl** (2026-01-14): https://x.com/ankrgyl/status/2011331850225271273 - Obvious: there's no excuse not to build a prototype anymore. Less obvious: there's no excuse not to build things the right way anymore. - **thdxr** (2026-01-13): https://x.com/thdxr/status/2011206781070913569 - @iamdavidhill OPENCODE_EXPERIMENTAL_DANGEROUSLY_USE_PLAN_MODE - **mikeyk** (2026-01-13): https://x.com/mikeyk/status/2011177881884639435 - There’s never been a better time to be a builder — Opus 4.5 & Claude Code keep surprising me in the quality and completeness of the products they can create. So I’m doing exactly that — putting my product founder hat back on and joining our Labs team to be hands-on at the frontier, building products that channel AI toward solving the world's hardest problems. Excited to pass the baton to Ami Vora as she leads the product team in scaling Claude. - **PlanetScale** (2026-01-13): https://x.com/PlanetScale/status/2011141183410671700 - "Systems were going down every day." Intercom's Ryan Sherlock walks through their entire database journey: MySQL, Aurora, ProxySQL, and why they ultimately chose PlanetScale. https://t.co/YlZdXHhbyg - **ritakozlov** (2026-01-13): https://x.com/ritakozlov/status/2011107126845063476 - everyone talking about how this is the end of entry level roles hasn’t worked with truly mid-level white collar management give me an AI-pilled junior person that knows how to deploy and army of agents and has no ego about it over a senior empire builder AI-luddite any day - **mattzcarey** (2026-01-13): https://x.com/mattzcarey/status/2011086975181234678 - elicitations in @opencode https://t.co/XNN36vG35O - **calebmer** (2026-01-13): https://x.com/calebmer/status/2010867840375795917 - The thing I've been building for 3 years is finally live https://t.co/OzXvrLq10D https://t.co/N90DObAL8j https://t.co/xSAaUw373O - **ritakozlov** (2026-01-12): https://x.com/ritakozlov/status/2010805467715670291 - code mode code mode code mode code mode code mode https://t.co/ttvJs4es2x - **mattzcarey** (2026-01-12): https://x.com/mattzcarey/status/2010781496630812688 - dropping this here, will talk more about it sometime this week. 2500 api endpoints now accessible from any agent the WHOLE of the fricking Cloudflare API The openapi spec alone is 2.3 MILLION tokens!! this uses only 1000 🔥 https://t.co/q11HLXKJWO - **zack_field** (2026-01-12): https://x.com/zack_field/status/2010748520878493878 - inspect is a blast to use. @zachbruggeman cooked. I pointed Inspect at our triage queue last week and it spun up 26 different sessions. only 8 of them produced a valid fix/improvement but ~30% of work being automated is... incredible. https://t.co/7wtPDkGC0Z - **ritakozlov** (2026-01-12): https://x.com/ritakozlov/status/2010736100810293572 - 2026 is the year of agents on @cloudflaredev agents sdk + durable objects https://t.co/rofPSvTtxt https://t.co/qRbGKNr6nY - **thdxr** (2026-01-12): https://x.com/thdxr/status/2010729452591054949 - it's insane how much ramp built - it's like our entire roadmap and they just went and built it for their team https://t.co/l3fAEcSRZd - **zachbruggeman** (2026-01-12): https://x.com/zachbruggeman/status/2010728444771074493 - The craft of engineering is rapidly changing. At @tryramp, we built our own background coding agent to accelerate faster. We call it Inspect. It wrote 30% of merged frontend + backend PRs in the past week. It’s powered by @opencode, @modal and @CloudflareDev. It runs fully in the cloud, and starts in seconds, letting every builder work at the speed of thought, no setup required. Today, we’re open sourcing the full blueprint so anybody can build their own Inspect. Just give our spec to your current coding agent, and let it build your new favourite. - **jarredsumner** (2026-01-12): https://x.com/jarredsumner/status/2010685702485254251 - github issues are prompts - **threepointone** (2026-01-12): https://x.com/threepointone/status/2010681124314358112 - yeah this is a good bet. wouldn't be surprised if we see success stories as early as q1 https://t.co/SdmchvL2BS - **jarredsumner** (2026-01-12): https://x.com/jarredsumner/status/2010675566664466606 - I think open source repos almost entirely maintained by LLMs will be a thing this year - **IrvingSwisher** (2026-01-12): https://x.com/IrvingSwisher/status/2010561285683830790 - Said this before and will say it again: The adverse impacts of undermining Fed independence are probably not going to show up in the stock market first, or most prominently. Wouldn't be shocked if the stock market all but ignored the events of this evening 24hrs from now. - **brezina** (2026-01-12): https://x.com/brezina/status/2010554496372633977 - @RoKhanna You do not have much credibility on anything related to money or finance after booting out the taxpayers that fund our state. ALL of our taxes are going to go up - **jaynordlinger** (2026-01-12): https://x.com/jaynordlinger/status/2010532518123200659 - Jerome Powell is like a ghost from our pre-2016 past. A sober, responsible, patriotic public official. His kind are nearing extinction. And isn’t it obvious why the Trump GOP would be hot to get rid of him? He’s the opposite of Trump-Vance-Hegseth-Noem-RFK-Gabbard-Patel … - **jasonfurman** (2026-01-12): https://x.com/jasonfurman/status/2010532384924442645 - Some countries that have prosecuted or threatened to prosecute central bankers for the purpose of political intimidation or punishment for monetary policy decisions: Argentina, Russia, Turkey, Venezuela and Zimbabwe. - **matt_slotnick** (2026-01-12): https://x.com/matt_slotnick/status/2010528305334948316 - days like today in SF remind me why if i had 10 billion dollars i’d gladly pay a one time $500m levy to avoid having to live in austin - **JustinWolfers** (2026-01-12): https://x.com/JustinWolfers/status/2010528264389800197 - What can history teach us about what happens when a populist strongman with an idiosyncratic taste for low interest rates undermines central bank independence? https://t.co/VvdaOq27ic - **SenAdamSchiff** (2026-01-12): https://x.com/SenAdamSchiff/status/2010527468361269409 - Horrified by the vile attack on the Beth Israel Congregation, and praying for Jackson, Mississippi’s Jewish community tonight. Across the country, we are seeing a disturbing rise in antisemitic hate speech and violence. It has no place in our society, and we must call it out in the strongest possible terms and ensure accountability. - **RajaKorman** (2026-01-12): https://x.com/RajaKorman/status/2010526094370410928 - The name is Bond. Bond Market. - **ianbremmer** (2026-01-12): https://x.com/ianbremmer/status/2010525101528400354 - banana republic action. independent fed chair reaction. https://t.co/aBZe0tlzZf - **JustinWolfers** (2026-01-12): https://x.com/JustinWolfers/status/2010524340639424564 - Attempting to criminalize the actions of an independent public servant, Fed Chair Jay Powell, for the sin of acting independently—in the public interest rather than the President's political interest—is an outrage. It's bad economics, bad politics, bad for the rule of law, bad for the public sector, bad for American credibility and bad for Americans. - **AdamWhitcroft** (2026-01-12): https://x.com/AdamWhitcroft/status/2010523334820397444 - Wanted a quick notes app so I made a coffee, opened terminal and hit up Claude. End result is a cmd+k editor that writes .md files in ~/Documents (collating if multiple per day) with basic formatting support. It'll ping me at 4pm weekdays for a quick eod summary. https://t.co/E4OYGBbBwE - **jasonfurman** (2026-01-12): https://x.com/jasonfurman/status/2010522125283795249 - A terrific statement from a true statesman. I am grateful for everything Chair Powell is doing to resist this outrageous attempt by the President to use lawfare to subvert the Fed’s responsibility to pursue the objectives set for it by law—maximum employment and price stability. https://t.co/dfSq5YjN96 - **pitdesi** (2026-01-12): https://x.com/pitdesi/status/2010521761427685413 - @thogge We have the #1 fraudster of all time in the Oval Office and he’s trying to put people in jail for fake fraud because he doesn’t like what they are doing. He’s pardoning actual fraudsters like Trevor Milton, George Santos, too many to remember at this pt https://t.co/1e26uiRHAm - **apartovi** (2026-01-12): https://x.com/apartovi/status/2010521346892087704 - I've spent the past week anxiously following the news in Iran. Protests in the last few days have escalated and taken a dark turn. Thousands have been killed or arrested, and the government has cut off phone and internet access, leaving tens of millions of people unable to communicate.  My heart goes out to the innocent people who are affected by all this, including Iranians who live in the US and elsewhere. ❤️ Many are attempting to continue their studies or work and get through their daily lives under the immense pressure of wondering what's happening in Iran. Some who are financially supported by their families in Iran are in need now. Some may have lost friends or family members. Many can’t even contact their family members to know if they’re safe. There are others who might have traveled home to Iran for Xmas and are now stuck, unable to get a flight back.   Even those who aren't directly impacted are silently carrying an emotional burden. I've spent much of the past week feeling helpless and longing for some way to help or be involved. I miss the city of my birth and feel the suffering of the people there.  If you're reading this and you work with anybody from Iran, be mindful of what they might be going through. Consider giving them space and understanding, extending deadlines, offering tuition relief.... or just a hug. Much love and hope. ❤️ - **zachklein** (2026-01-12): https://x.com/zachklein/status/2010521101185499371 - Incredible. https://t.co/rlYXB8SQK0 - **soleio** (2026-01-12): https://x.com/soleio/status/2010519320850510120 - Can we just take a moment to appreciate how insane this throw on the run was by a wide receiver and the hit he took right after? Jennings was so clutch here. https://t.co/joevQQr4rh - **HighyieldHarry** (2026-01-12): https://x.com/HighyieldHarry/status/2010516717814894905 - Can't believe this has to be said, but you have to be a fucking moron to think Powell and the Fed are actually guilty of some sort of building renovation conspiracy. - **M_C_Klein** (2026-01-12): https://x.com/M_C_Klein/status/2010516325005930531 - 👀 "I will oppose the confirmation of any nominee for the Fed—including the upcoming Fed Chair vacancy—until this legal matter is fully resolved." https://t.co/ctkzBGkdpY - **McFaul** (2026-01-12): https://x.com/McFaul/status/2010516266788798596 - This is crazy. Republicans need to denounce this attack on Powell and the Fed with passion. We cannot allow Trump to destroy our democracy. Republican leaders, our Republic needs you now. https://t.co/YioObuoZGV - **federalreserve** (2026-01-12): https://x.com/federalreserve/status/2010510130970849338 - Video message from Federal Reserve Chair Jerome H. Powell: https://t.co/5dfrkByGyX https://t.co/O4ecNaYaGH - **ryancarson** (2026-01-11): https://x.com/ryancarson/status/2010467429177237534 - Another amazing way to use Ralph is to complete tons of user testing in the browser. I create a list of 10-20 detailed user testing scenarios (I specify that all acceptance criteria must be able to be judged by Amp using Chrome Dev Tools vs a human) and then I put them in a Ralph loop with Amp. Waaaay more testing than I would’ve ever done by hand. Then at the end I ask Amp to review progress.txt and tell me how it went. - **addyosmani** (2026-01-11): https://x.com/addyosmani/status/2010440624567460290 - Chrome DevTools now supports *individual* network request throttling! Folks have been asking for this for years! DevTools now allows developers to simulate slow network conditions for specific requests rather than the entire page. This helps in testing how a web application performs and handles issues when specific resources (like images, scripts, or API calls) are slow to load. - **dillon_mulroy** (2026-01-11): https://x.com/dillon_mulroy/status/2010411941555323172 - i've been following this pattern of creating skills and then wrapping them with commands like so and it's been working pretty well https://t.co/U7EPMelL0G - **danshipper** (2026-01-11): https://x.com/danshipper/status/2010411900861813233 - introducing proof! it's an agent-native markdown editor with provenance tracking. keeps track of what was written by you vs. AI, and lets you mark which parts of a plan document (or any .md file) you've reviewed, and how closely you've reviewed it. red = written by AI, no review. yellow = i've skimmed it green = i endorse this / i wrote this i think it'll be great as the default plan editor in Claude Code and Codex. releasing internally @every this week and if it goes well, will release more broadly - **faxherbert** (2026-01-11): https://x.com/faxherbert/status/2010396646576656548 - The @tryramp engineering team built a coding agent last month called Inspect and now it’s dominating our internal use cases. I can’t tell you how fun it is to continually have my mind blown by our technical folks. https://t.co/megpZXGjju - **Polymarket** (2026-01-11): https://x.com/Polymarket/status/2010390406056566967 - BREAKING: Man scales Iranian embassy in London, tears down the flag of the Iranian regime to raise the flag of Lion and Sun. https://t.co/ulabH3QTyZ - **dillon_mulroy** (2026-01-11): https://x.com/dillon_mulroy/status/2010387109748883749 - with a cloudflare theme https://t.co/QtQtSrw6iq https://t.co/IVr9CftcVt - **FredKSchott** (2026-01-11): https://x.com/FredKSchott/status/2010360577995272400 - Don’t forget that Tony Stark built Jarvis himself. Jarvis wasn’t SaaS. https://t.co/j5a2m3skds - **ianbremmer** (2026-01-11): https://x.com/ianbremmer/status/2010359187449012442 - us college activists would gain a little credibility if they were demonstrating against the brutal repression in iran. - **AndrewYang** (2026-01-11): https://x.com/AndrewYang/status/2010344044090831291 - I think the US should help the people of Iran. - **shervin** (2026-01-11): https://x.com/shervin/status/2010335841202430060 - From 2022 when there were only 9 Starlinks in Iran and a group of us begin to smuggle the first Starlinks into Iran to get it to the young and courageous rebels. https://t.co/QwYvm2UGvA - **TheCinesthetic** (2026-01-11): https://x.com/TheCinesthetic/status/2010304429233463543 - Begin Again (2013) is one of those rare feel-good films that deserves way more love and attention. https://t.co/vEoMT4Ctrw https://t.co/JApmJb3r3d - **docmilanfar** (2026-01-11): https://x.com/docmilanfar/status/2010234495979659299 - Iranian women - always at the forefront - protesting the forced hijab in 1979, shortly after the takeover by the ayatollah. https://t.co/jBvHimVgsm - **dillon_mulroy** (2026-01-11): https://x.com/dillon_mulroy/status/2010233421092794846 - concept https://t.co/6xfDhgAeBG https://t.co/ELI3HJmxr3 - **adamdotdev** (2026-01-11): https://x.com/adamdotdev/status/2010152035388195258 - In the early OpenCode days I drove myself crazy watching our competitors. I’m a competitive guy (whole anomaly team is), and there are a ton of teams working on coding agents. Then a couple months ago I muted them all lol; the individuals, the terms, the org accounts, all of it. I just interact with our users on Discord and build. It feels so good, complete 180 for my mental health. Now it’s like we don’t have any competition, and I’m not even sure if it’s because I muted it all or because we actually won. But it feels great either way. - **YIMBYLAND** (2026-01-10): https://x.com/YIMBYLAND/status/2010056394503004250 - FREE IRAN FREE IRANIAN BRICK MASONS https://t.co/J3XGheAFub - **zeeg** (2026-01-10): https://x.com/zeeg/status/2009885405269692576 - Something we've been trying out the last few days: instead of prompt stuffing https://t.co/ZYGe8ycXga (especially in subfolders), add skills to the repository. So far the results seem better. Its totally possible its entirely behavioral ("just tell it to use the skill" vs assuming it does the right thing), but thats actually fine. Two things that we've done that I think are successful: 1) A bunch of tedium has gone into global skills (https://t.co/8Ok6YObi0d) 2) Specific skills in repos that follow patterns ("ui-design", "db-migration", "write-docs", etc.. https://t.co/4P1pZqtDq3) Then instead of throwing everything into https://t.co/ZYGe8ycXga, its a lot simpler to just refer it to skills ("when doing X, use Y skill). It requires you to treat the skills like you'd treat any other code investment: maintain and distribute them. This is why I think its also important to do project-scoped configuration for everything (note: see the claude-settings-audit skill). You really need to build these once for the team (or once for the project), and maintain and iterate on them when they have problems. Anyways, so far so good. Nothing ground breaking - just an easy way to share prompt templates - but its smoothing out a lot of silly time sinks. I really want to see some of these behaviors become more cross-agent capable. Cursor's agent will pick up the in-repo skills, but the Claude plugins are specific to Claude right now. - **theo** (2026-01-10): https://x.com/theo/status/2009847953926893594 - Anthropic handed Anomaly/OpenCode the wildest W I've ever seen in the dev world. Truly insane. https://t.co/pVN50WKeao - **thdxr** (2026-01-10): https://x.com/thdxr/status/2009846100841160789 - this is coming too https://t.co/PQ4YwIU9pc - **dillon_mulroy** (2026-01-10): https://x.com/dillon_mulroy/status/2009839779983061149 - @ChristoPy_ OCaml-fying typescript* - **ritakozlov** (2026-01-10): https://x.com/ritakozlov/status/2009815251278279166 - rooting for the people of iran! incredible to watch the courage of people standing up to an oppressive, murderous regime https://t.co/xP5hXMHKdK - **jayair** (2026-01-10): https://x.com/jayair/status/2009807944087687411 - Think about where we started today and look at how it ended From people thinking we were done to all time highs and a new position You play the game for days like this OpenCode is now officially here https://t.co/njEsm2DbQO - **opencode** (2026-01-10): https://x.com/opencode/status/2009805930377167233 - https://t.co/E7VIHIfmeV https://t.co/D4KTJ0WHMC - **thdxr** (2026-01-10): https://x.com/thdxr/status/2009803906461905202 - in opencode v1.1.11 you can now use your ChatGPT Plus/Pro plans in OpenCode /connect to set it up https://t.co/jgiDNbRGrz - **tobi** (2026-01-10): https://x.com/tobi/status/2009788652218695727 - at least for small tools, keeping the code and throwing away the prompts is the 2025 equivalent of throwing away the source and keeping the binary. - **adamdotdev** (2026-01-10): https://x.com/adamdotdev/status/2009783076185063897 - Whole bunch of devs about to find out that GPT-5.2 is really good at programming - **gdb** (2026-01-09): https://x.com/gdb/status/2009776399893070004 - ❤️ https://t.co/DfWLEFB3hB - **theo** (2026-01-09): https://x.com/theo/status/2009763341624074692 - This is why I still think OpenAI is going to win. Restrictions win short term. Collaboration wins long term. https://t.co/lMpp8gd9QQ - **elithrar** (2026-01-09): https://x.com/elithrar/status/2009756446804549770 - big "I told you so" moment for @dok2001 when he pushed an @opencode command + subagent to our central config that runs three code review sub-agents in parallel + reports back. wasn't convinced it was going to be better than a single one, but I was wrong. temp 0.1 helps a ton. https://t.co/LxODu9lJId - **dkundel** (2026-01-09): https://x.com/dkundel/status/2009747124628639859 - For all you OpenCode fans 💖 https://t.co/AAs1RFuzeG - **thdxr** (2026-01-09): https://x.com/thdxr/status/2009744706314403868 - guess we're going to be working hard to make opencode work as best as possible with gpt5 - **thdxr** (2026-01-09): https://x.com/thdxr/status/2009742787323834805 - https://t.co/nlB7GVnZEj - **thsottiaux** (2026-01-09): https://x.com/thsottiaux/status/2009742187484065881 - We are working with OpenCode to allow Codex users to use their Codex subscriptions and usage limits in OpenCode directly. Also exploring how to support other awesome actors in the space. https://t.co/Wl9pCHNFsA - **thdxr** (2026-01-09): https://x.com/thdxr/status/2009742070471082006 - we are working with openai to allow codex users to benefit from their subscription directly within OpenCode - **jamwt** (2026-01-09): https://x.com/jamwt/status/2009737122039902606 - 1. The models are commodities now, but not the tools. So use things like opencode and t3 chat. Let the labs compete brutally to your benefit. 2. Expect to eventually pay $1000-1500 per dev once the loss leadering is done (capital will wane eventually) 3. For most, still worth it - **thsottiaux** (2026-01-09): https://x.com/thsottiaux/status/2009714843587342393 - With the Claude Code shutdown. I am proud that we build Codex in the open with our OSS repo and we are 100% invested in supporting a flourishing ecosystem of agentic coding tools out there. You can already build on top of https://t.co/hDjdmjH8pg directly, which includes ChatGPT login and same usage as you get in codex. Reach out if you are a builder! - **jshchnz** (2026-01-09): https://x.com/jshchnz/status/2009691791482859539 - No more excuses to not help fund the open source software that powers your code Just type /tribute in Claude Code https://t.co/4lpoeib7d1 - **DeTahmineh** (2026-01-09): https://x.com/DeTahmineh/status/2009680255091405074 - **eastdakota** (2026-01-09): https://x.com/eastdakota/status/2009654937303896492 - Yesterday a quasi-judicial body in Italy fined @Cloudflare $17 million for failing to go along with their scheme to censor the Internet. The scheme, which even the EU has called concerning, required us within a mere 30 minutes of notification to fully censor from the Internet any sites a shadowy cabal of European media elites deemed against their interests. No judicial oversight. No due process. No appeal. No transparency. It required us to not just remove customers, but also censor our 1.1.1.1 DNS resolver meaning it risked blacking out any site on the Internet. And it required us not just to censor the content in Italy but globally. In other words, Italy insists a shadowy, European media cabal should be able to dictate what is and is not allowed online. That, of course, is DISGUSTING and even before yesterday’s fine we had multiple legal challenges pending against the underlying scheme. We, of course, will now fight the unjust fine. Not just because it’s wrong for us but because it is wrong for democratic values. In addition, we are considering the following actions: 1) discontinuing the millions of dollars in pro bono cyber security services we are providing the upcoming Milano-Cortina Olympics; 2) discontinuing Cloudflare’s Free cyber security services for any Italy-based users; 3) removing all servers from Italian cities; and 4) terminating all plans to build an Italian Cloudflare office or make any investments in the country. Play stupid games, win stupid prizes. While there are things I would handle differently than the current U.S. administration, I appreciate @JDVance taking a leadership role in recognizing this type of regulation is a fundamental unfair trade issue that also threatens democratic values. And in this case @ElonMusk is right: #FreeSpeech is critical and under attack from an out-of-touch cabal of very disturbed European policy makers. I will be in DC first thing next week to discuss this with U.S. administration officials and I’ll be meeting with the IOC in Lausanne shortly after to outline the risk to the Olympic Games if @Cloudflare withdraws our cyber security protection. In the meantime, we remain happy to discuss this with Italian government officials who, so far, have been unwilling to engage beyond issuing fines. We believe Italy, like all countries, has a right to regulate the content on networks inside its borders. But they must do so following the Rule of Law and principles of Due Process. And Italy certainly has no right to regulate what is and is not allowed on the Internet in the United States, the United Kingdom, Canada, China, Brazil, India or anywhere outside its borders. THIS IS AN IMPORTANT FIGHT AND WE WILL WIN!!! - **dillon_mulroy** (2026-01-09): https://x.com/dillon_mulroy/status/2009653238728204345 - introducing better-result, my opinionated take on the rust result type for typescript after having written about 100 times npm i better-result *disclaimer*: if you can use effect-ts, you should use effect-ts https://t.co/D79J6GMPmp - **danshipper** (2026-01-09): https://x.com/danshipper/status/2009651408144835021 - NEW: i wrote a complete technical guide to building agent-native software (co-authored with claude) it covers: - the five pillars of agent native design (parity, granularity, composability, emergent capability, self-improvement) - files as the universal interface - agent execution patterns with code samples - mobile agent patterns - advanced patterns like dynamic capability discovery if you want to take full advantage of this moment, it's worth your time: https://t.co/IYTnXjYScO - **yomnashousha** (2026-01-09): https://x.com/yomnashousha/status/2009649986255139194 - Get automatically notified when your @Cloudflare Workers Builds succeed or fail - Generates prod/preview URLs when builds succeed - Receive error message when builds fail - Send notifications to Slack, Discord, orany webhook destination! Get started 👇 https://t.co/IJ6XmqqTLT - **dillon_mulroy** (2026-01-09): https://x.com/dillon_mulroy/status/2009619822959632782 - gm to everyone doing the same thing today https://t.co/GzvSBQp8ms - **thdxr** (2026-01-09): https://x.com/thdxr/status/2009618812631503085 - on the anthropic stuff (technical details in a reply) it's their business they have the right to enforce their terms however they like they're not obligated to provide completely open access to their services but it is a hint at the underlying problem - models aren't sticky so it's not good enough that you're using opus they need you and your team to adopt their full stack of tooling so that it is hard to switch to a better model thankfully not all LLM providers want to operate this way - you'll hear more about that soon :) in the meantime we already teased more OpenCode Black dropping today - so stay tuned - **iamdavidhill** (2026-01-09): https://x.com/iamdavidhill/status/2009536853335011564 - i’ve never been more bullish on open source - **bcherny** (2026-01-09): https://x.com/bcherny/status/2009450715081789767 - We just open sourced the code-simplifier agent we use on the Claude Code team. Try it: claude plugin install code-simplifier Or from within a session: /plugin marketplace update claude-plugins-official /plugin install code-simplifier Ask Claude to use the code simplifier agent at the end of a long coding session, or to clean up complex PRs. Let us know what you think! - **mitsuhiko** (2026-01-08): https://x.com/mitsuhiko/status/2009368000827195510 - I mean honestly this is pretty fucking cool. I really only put the agent to an API and now I can ask it questions that the web interface absolutely cannot answer. https://t.co/8YW5xaurTi - **thdxr** (2026-01-08): https://x.com/thdxr/status/2009359614975025495 - crazy milestone that omarchy is now shipping opencode by default if coding agents become an important tool for building software they can't all be proprietary and vendor locked it's not that opencode is amazing yet, there's just a deep need for a project like it https://t.co/A8xqSop29A - **TheAppleDesign** (2026-01-08): https://x.com/TheAppleDesign/status/2009355748124635194 - Maybe it’s just me but i miss MacBook battery indicators.. such an underrated feature https://t.co/icj3U6Y5ey - **fnthawar** (2026-01-08): https://x.com/fnthawar/status/2009326676719886390 - new reality I spent a lot of August (Claude Code) and December (Antigravity) coding... so much has changed Build and learn what's possible https://t.co/NKdECW10Ze - **joshpuckett** (2026-01-08): https://x.com/joshpuckett/status/2009126748093657467 - what if claude code auto-compactor felt like this... https://t.co/bTC5Z9v13o - **KentonVarda** (2026-01-08): https://x.com/KentonVarda/status/2009114611312951784 - TL;DR: This optimization makes it so that when a large HTTP response (or request) body is proxying through Workers, we use the Linux splice() syscall to move the bytes from socket to socket, so they never even enter userspace. - **CSElmendorf** (2026-01-08): https://x.com/CSElmendorf/status/2009109866321989788 - Tl, dr: City of Mountain View charges developers $360k/unit for the privilege of building townhomes. https://t.co/amg3GiifRG - **thdxr** (2026-01-08): https://x.com/thdxr/status/2009109485219094551 - a lot of people try to contribute big features to opencode which are unlikely go through given they need a lot of product thought would love it instead if people identified bad code and cleaned it up - easy merge, low coordination needed here's an example https://t.co/0nBy5B71JR - **brian_lovin** (2026-01-07): https://x.com/brian_lovin/status/2009046532218372262 - This post made me realize I don't actually understand how terminal UIs work. I asked Claude to make an explainer with interactive examples. https://t.co/uFM00fIcqt It still feels like black magic, but now it's slightly less magic. - **jonas** (2026-01-07): https://x.com/jonas/status/2009035600465105272 - Please join this great team! We need more people working on the infra and platform layer below all the fancy SDKs ;) https://t.co/HaC2qu3gKJ - **CoachDanCasey** (2026-01-07): https://x.com/CoachDanCasey/status/2009025578083733981 - Christian McCaffery trains like a Sprinter. There’s a reason you should run Track. https://t.co/7MyiiT6Ney - **elithrar** (2026-01-07): https://x.com/elithrar/status/2009022327753973811 - If you're a seasoned distributed systems engineer, understand the Actor model, know C++ and want to work on a system that not only underpins some of our largest customers' apps but also many of our own, then DM me. The @cloudflare Durable Objects team is hiring. - **ivanhzhao** (2026-01-07): https://x.com/ivanhzhao/status/2009022162611450327 - @adamdangelo Assembly code! - **nateberkopec** (2026-01-07): https://x.com/nateberkopec/status/2009011736385605990 - OpenCode has a few killer features, and this is why I'm switching: 1) Proper queued message support, jesus christ Claude Code get it together 2) Don't have to switch back/forth between TUIs for Codex/Claude 3) /share is extremely useful It's also just pretty! - **hraness** (2026-01-07): https://x.com/hraness/status/2008992184108785702 - The big shift for me recently has been realizing I should skip the todo, and just write the prompt. No more Linear ticket. Now, simple tasks go straight to pull request. I've started using @zocomputer to orchestrate Claude Code agents for these small tasks. Building Zo with Zo! And often, I'm on the go. I can just text my Zo to kickoff a task. It's so awesome. - **gdb** (2026-01-07): https://x.com/gdb/status/2008984705723719884 - Introducing ChatGPT Health: https://t.co/HnzAoH6UtV https://t.co/qLYFSAxyZW - **akaphill** (2026-01-07): https://x.com/akaphill/status/2008965919486431421 - Weft. I wanted a personal board where agents could pick up some of my daily tasks. Think Trello, but agents do your tasks for you. Agents can connect to Gmail, Docs, GitHub, or any MCP server. All tasks have access to a Sandbox with Claude Code. Built on @CloudflareDev https://t.co/eGo2KMmeR1 - **adamdangelo** (2026-01-07): https://x.com/adamdangelo/status/2008961539836186773 - i can't believe we used to write all this code by hand - **hraness** (2026-01-07): https://x.com/hraness/status/2008956003643986157 - https://t.co/CqV8AwrVWU - **ankrgyl** (2026-01-07): https://x.com/ankrgyl/status/2008950530077544938 - trace lineup just dropped 👀 https://t.co/DO7IKwB9rw - **DanielNealAdler** (2026-01-07): https://x.com/DanielNealAdler/status/2008947942728847737 - Introducing tin, an open-source VCS for AI coding conversations, and my holiday hack project https://t.co/2oAfANTn6L Today, all real design work happens in conversations with AI agents, and those conversations disappear into local caches. tin turns those threads into first-class, version-controlled artifacts, permanently linked to the code they produced. tin retains the human thinking behind the code, the history of decisions that led to the code, and the provenance of the code. It provides a structural linkage between agent conversations and code changes. - **AnjneyMidha** (2026-01-07): https://x.com/AnjneyMidha/status/2008944401100546050 - when i was ceo of my first startup, i made every mistake imaginable. but the most egregious one was doubting my own intuition. that one is the cardinal sin. - **4AAAAart** (2026-01-07): https://x.com/4AAAAart/status/2008937669330407907 - A Topology Of Californian Topiary Photography by Marc Alcock https://t.co/6yeLBtIuKA - **adamdotdev** (2026-01-07): https://x.com/adamdotdev/status/2008922657907995023 - Side note: the number of QoL fixes and stuff coming in from outside contributors at this point is incredible, I cannot imagine being one of our closed source competitors, sounds absolutely awful https://t.co/8pxDIXvNLz - **threepointone** (2026-01-07): https://x.com/threepointone/status/2008914621168587013 - opencode feature request: (or maybe rfc for a new plugin) - take a share link and _continue_ hacking on it with a sandbox and dev envt behind it. bonus: if I can merge that work back to the request (with modifications to the PR description). also how would that work in a team? https://t.co/HXV0ncPzWf - **threepointone** (2026-01-07): https://x.com/threepointone/status/2008913942555336714 - new post: the context is the work https://t.co/oCPnFHh7mu (aka write better PR descriptions) wherein I argue the specifics of how everything before you say "ok write the code" is the actual work. also more juniors, onboarding, and missing features from github et al. - **schrockn** (2026-01-07): https://x.com/schrockn/status/2008894863597883869 - Current status: taxiing in a plane, tethered to my phone, launching claude code plans to execute in github actions (have custom tools for this) so that they are done and I can review when the wifi turns back on. Coding in 2026 is so goddamned wild. - **doodlestein** (2026-01-07): https://x.com/doodlestein/status/2008813776687030781 - What makes for a good markdown plan document for a software development project? What's the difference between a good plan and a great one? I'm always going on and on about how I spend 85%+ of my time and energy on the planning phases. Well what does that entail exactly? It's hard to explain in the abstract; you need a tangible example to really illustrate the nuances. So I thought I would share a good example from today. This also addresses a recent question I've been getting about my approach. People seem to be under the impression that you have to do everything in your project in one shot. And in my approach, that is true, but only for version 1! If you decide you want to add new features or change how things work, you can obviously do that once you have a functioning v1. And the way you do that is the same way you create v1, by creating a super detailed markdown plan first and then turning it into beads. So I'll give you an example from Cass, my Coding Agent Session Search program, which is a pretty elaborate Rust program that automatically detects, parses, stores, and indexes all your previous session logs from just about every coding agent out there. It offers sub 50ms instant "search as you type" across all those logs and has many other nice features. I decided that I'd like to add a feature to Cass that is similar to a feature I already have in MCP Agent Mail and in beads_viewer (bv): the ability to export your setup as a static website that can be served using GitHub Pages. You can see an example for bv for this very project, which is the final result of the planning process I will describe in this post: https://t.co/iw50xoFBgd This functionality makes it very quick and easy to generate and deploy the exported site using the gh utility. The site itself usually consists of an sqlite file and a bunch of typescript and wasm that runs entirely in the browser, but with very good performance and nice features and styling, which you can observe in the example just given. Now, sharing MCP Agent Mail messages or a bunch of beads is one thing, but sharing a bunch of coding agent session logs is very different; these things are often full of sensitive information, API keys, curses/invective (at least mine are!), and other material that you definitely wouldn't want to expose to the world. But GitHub Pages, as nice as it is, only works for public repos (btw, my tools also support Cloudflare pages, but GH Pages is better and easier for this use case). So how to handle these problems? The answer is encryption: the user first selects which coding agents to include, which project folders, the time period, etc. and a bundle is generated (note that this bundle is in the canonical format that Cass internally converts all coding agent messages into from their original native formats) and then the user provides a password to use for the encryption of that bundle. So the idea is that although the repo and web page are public, everyone but you and others you tell the password to will simply see a password field and will be unable to read any of the messages. Once the password is entered, it would unlock a beautiful, responsive UI that lets you easily search across the messages nearly as quickly and efficiently as Cass does. And if you really have nothing to hide, you can leave the password off and make it all actually public. Anyway, this is a super complex new feature to add for many reasons. The export wizard alone is extremely intricate and complex. But the additional requirements around encryption and security make this hard to get right. Anyway, I asked Claude Code with Opus 4.5 to first study all the relevant code from my bv project for implementing this feature, and then explained how the corresponding feature in Cass would have to differ. It then created for me an initial cut at a plan. I asked CC to check over the bv code again for more insights and understanding into how the feature was implemented in bv and to then use those insights to expand and improve the plan document in-place. Finally, I moved the plan over to ChatGPT Pro in the browser with GPT 5.2 with Extended Reasoning along with this prompt preamble: "Carefully review this entire plan for me and come up with your best revisions in terms of better architecture, new features, changed features, etc. to make it better, more robust/reliable, more performant, more compelling/useful, etc. For each proposed change, give me your detailed analysis and rationale/justification for why it would make the project better along with the git-diff style change versus the original plan shown below:" I then pasted the outputs into Claude Code like this: OK, now integrate these revisions to the markdown plan in-place; use ultrathink and be meticulous. At the end, you can tell me which changes you wholeheartedly agree with, which you somewhat agree with, and which you disagree with: ```[Pasted text #1 +995 lines]``` I then rinsed and repeated this another 2 times. This is not a fast process: the last review, which you can see a screenshot of below, took 27 minutes to finish: You can see the final ~3,500-line markdown plan here: https://t.co/tez9ZfVpjc But what's more interesting is to see the revisions in GitHub, where you can see the massive improvements that are made from v2 to v3, and how the improvements actually continue all the way to the end: This ended up taking a total of around 3 hours. I suspect this is why most people seem to be unwilling to do this level of planning. It feels like you're not getting much done because no code is being written, but if you do it correctly and then start up enough agents in your swarm with Agent Mail, Beads, and bv, then the code will be written so ridiculously quickly that it more than makes up for this slow part. And what's more, the code will be really good. Returning to the story, I was then ready to turn the plan into beads. I've already explained that process in another recent post, but the essence of it is to use this prompt with Claude Code: "OK so now read ALL of PLAN_TO_CREATE_GH_PAGES_WEB_EXPORT_APP.md; please take ALL of that and elaborate on it and use it to create a comprehensive and granular set of beads for all this with tasks, subtasks, and dependency structure overlaid, with detailed comments so that the whole thing is totally self-contained and self-documenting (including relevant background, reasoning/justification, considerations, etc.-- anything we'd want our "future self" to know about the goals and intentions and thought process and how it serves the over-arching goals of the project.). The beads should be so detailed that we never need to consult back to the original markdown plan document. Remember to ONLY use the `bd` tool to create and modify the beads and add the dependencies. Use ultrathink." Then I just did round after round of this follow-up prompt in Claude Code: "Reread AGENTS dot md so it's still fresh in your mind. Then read ALL of PLAN_TO_CREATE_GH_PAGES_WEB_EXPORT_APP.md . Use ultrathink. Check over each bead super carefully-- are you sure it makes sense? Is it optimal? Could we change anything to make the system work better for users? If so, revise the beads. It's a lot easier and faster to operate in "plan space" before we start implementing these things! DO NOT OVERSIMPLIFY THINGS! DO NOT LOSE ANY FEATURES OR FUNCTIONALITY! Also make sure that as part of the beads we include comprehensive unit tests and e2e test scripts with great, detailed logging so we can be sure that everything is working perfectly after implementation. It's critical that EVERYTHING from the markdown plan be embedded into the beads so that we never need to refer back to the markdown plan and we don't lose any important context or ideas or insights into the new features planned and why we are making them." After around 8 or 9 rounds, it finally reaches a steady state. I then had Codex with GPT 5.2 high reasoning effort do one final round using that same prompt. The final result can be seen at the link shared above. And that, friends, is what a good plan looks like. And also what a good set of beads based on a plan looks like. Those beads are so polished and detailed that you can mechanically unleash a big swarm of agents to implement it using Agent Mail, Beads, and bv, and it will come out just about perfectly without much fuss. I'm too tired to start doing that implementation work tonight, but I'll do it tomorrow and will then share the new feature, which I think Cass users are going to really like. - **thdxr** (2026-01-07): https://x.com/thdxr/status/2008691230339645457 - opencode's server right now is a great way to embed a coding agent anywhere and talk to it its next iteration will let you orchestrate many agents running anywhere - **schrockn** (2026-01-06): https://x.com/schrockn/status/2008658515406320004 - The dev workflow for github actions is where productivity goes to die. - **elithrar** (2026-01-06): https://x.com/elithrar/status/2008657921593536771 - what about a pattern where --help returns JSON of the entire command surface + flags + descriptions when invoked from a non-interactive shell (via an agent) to help guide it more comprehensively? <cli> create-skill to output an agent skill? https://t.co/qfmre3TE06 - **samgoodwin89** (2026-01-06): https://x.com/samgoodwin89/status/2008645252853493822 - People think I am taking a risk by going all in on Effect. Oh no, what will the users think? Big scary Effect 😱 Bro, Claude is the user now. Claude loves Effect. Developer preference used to matter. Not anymore. - **thdxr** (2026-01-06): https://x.com/thdxr/status/2008624441106772183 - OpenCode Black use any model $200/month limited run for now, link to buy in reply https://t.co/YYdBZzaATC - **opencode** (2026-01-06): https://x.com/opencode/status/2008624165020979339 - OpenCode Black Use any model $200/month Limited run for now Link in reply https://t.co/TtSV2RselO - **heydna_** (2026-01-06): https://x.com/heydna_/status/2008623696143905057 - hey hi I'm Deanna and I'm the PM for @Cloudflare Images what do you want to see in Images? 👀 what's missing? what would make your life better? https://t.co/wEuQ2jd3FH - **mattpocockuk** (2026-01-06): https://x.com/mattpocockuk/status/2008622410342261107 - A Ralph script for writing tests on untested features: https://t.co/aUPkInl5aN - **thdxr** (2026-01-06): https://x.com/thdxr/status/2008564628096491559 - we don't hear from adam for one night and then we end up with this https://t.co/92ArsrGsjs - **ryancarson** (2026-01-06): https://x.com/ryancarson/status/2008548371712135632 - https://t.co/b8RRfADhx5 - **ritakozlov** (2026-01-06): https://x.com/ritakozlov/status/2008542325006532780 - for the first time, theory is starting to feel closer to reality 2026 is the year of automating yourself https://t.co/K26oP2d477 - **zebassembly** (2026-01-06): https://x.com/zebassembly/status/2008540765891555370 - My team is hiring! We're looking for a full-stack engineer that can help us make observability for Workers better. If you're in the austin/london/lisbon area and looking to build the Cloudflare developer platform on the Cloudflare developer platform please check the replies! - **dillon_mulroy** (2026-01-06): https://x.com/dillon_mulroy/status/2008400880119296500 - this. something like this please https://t.co/dXoCiyTQ1n - **hudzah** (2026-01-06): https://x.com/hudzah/status/2008352007174697425 - a mockup: doing PR reviews by seeing a programmers chain of prompts! it also encourages devs to be deliberate with how they instruct llms as they become more capable, and poor human instructions + context ends up being the reason why code doesn't work https://t.co/nrUv8OmF7N - **dillon_mulroy** (2026-01-06): https://x.com/dillon_mulroy/status/2008348161882882166 - @samgoodwin89 don't do ralph dirty like that https://t.co/S7T8l2kqyY - **heyeaslo** (2026-01-06): https://x.com/heyeaslo/status/2008331102138564976 - This is the Xteink X4, and I’m adding it to my curated tech list. https://t.co/Siey4tDOhC - **mdo** (2026-01-05): https://x.com/mdo/status/2008312716050747403 - Let me introduce y’all to Diffs—a new, open source diff and code rendering library. We’ve been working on it @pierrecomputer for a few months now and we’re stoked for you to see it. https://t.co/G0ehctDaiY - **realdavidleen** (2026-01-05): https://x.com/realdavidleen/status/2008299651372585427 - One of my projects from over the break: https://t.co/jPllwoSxr0 1. An MCP server hosted on Cloudflare, 2. Which supports dynamic client registration via Auth0 so you can use it with claude code authentication, 3. Which can read your Google Calendar using access tokens, 4. The access tokens are retrieved via Auth0 token vault using token exchange, 5. You connect your Google Calendar account to Auth0 using a single page app for the consent screen, also hosted on Cloudflare, 6. Escalating to push notifications for approval using the Auth0 Guardian app (as an example) for sensitive operations e.g. deleting an event on the calendar This project answered a lot of questions for me as to how a background/async agent could conceptually integrate at a large enterprise while maintaining a strong security posture. It's likely quite tricky to deploy as even though I tried stick to infra-as-code some things can only be done via the Auth0 console. Many thanks to Claude Code for patience and interesting discussions. @trq212 for interesting projects built over the holidays. @yenkel and @whoiskatrin - you might find this interesting. - **Esteban_Puerta9** (2026-01-05): https://x.com/Esteban_Puerta9/status/2008282881370554745 - https://t.co/NprWWsjoIf - **swyx** (2026-01-05): https://x.com/swyx/status/2008274475301871834 - turns out ai-augmented understanding is useful for a lot more than just codemaps. got a preview of @walden_yan's next Devin update and am pretty hyped; look out for it next week https://t.co/JpNfyBqVQ4 - **GergelyOrosz** (2026-01-05): https://x.com/GergelyOrosz/status/2008272970041409579 - Unpopular option: most change that AI tools will bring for software engineers are likely to be making the practices that the best eng teams did until now, the baseline for those that want to stay competitive + move fast Things like product-minded engineers, testing, o11y, CD etc - **michellechen** (2026-01-05): https://x.com/michellechen/status/2008246972059447409 - built a small github bot that compiles LaTex files → outputs pdf https://t.co/yNuXjQl71h fresh PDFs on every PR — collaborate on papers, update resumes, write problem sets, etc. no latex install needed, fully serverless. built with @CloudflareDev sandbox sdk - **dbreunig** (2026-01-05): https://x.com/dbreunig/status/2008246535268741494 - At our Bay Area DSPy meetup, @lateinteraction delivered an incredible talk about the philosophy of DSPy, why it matters, and treating AI engineering as an actual engineering discipline. Can't recommend this one enough: https://t.co/TUKnqBhEAW - **dillon_mulroy** (2026-01-05): https://x.com/dillon_mulroy/status/2008244182952145028 - @ericclemmons @Cloudflare @CloudflareDev @alchemy_run @threepointone so stoked! welcome!!!!! - **ericclemmons** (2026-01-05): https://x.com/ericclemmons/status/2008239725950497206 - 🥳 I'm joining @Cloudflare Jan 20th here in Austin! @CloudflareDev has been on 🔥 for a while and I've been using Cloudflare for *everything* for the past ~year (with @alchemy_run) Friends were pushing me to join ("It's a natural fit!") I'm glad I listened 🙏 https://t.co/g6G3GmBnYp - **willccbb** (2026-01-05): https://x.com/willccbb/status/2008228531067908334 - Ralph Wiggum is the craziest thing to happen in the coding agents space in 2026 so far not the technique itself. just the fact that someone was like "here's a technique it's called the Ralph Wiggum technique because of the Simpsons guy" and everyone was like "OK sounds good" https://t.co/ZebK15k9H2 - **burcs** (2026-01-05): https://x.com/burcs/status/2008205188113518659 - been dreaming about this for a while… imagine a personal agent that controls your mac from anywhere, chat to it from your phone, it takes action on your machine built with cloudflare agents + raycast https://t.co/F4hLvbHhKw - **mattpocockuk** (2026-01-05): https://x.com/mattpocockuk/status/2008142709526344117 - How it started: Swarms, Orchestrators, Agent meshes, Multi-agent frameworks, Hierarchical agent systems, Agent-to-agent protocols How it's going: https://t.co/58plOaLUkr - **swyx** (2026-01-05): https://x.com/swyx/status/2008079127538983173 - this might be the fastest way to tune the x feed rn (settings -> timeline -> post interaction) default twitter makes it very hard/inconvenient to downrank posts (need 2 clicks w very small surface area). now can just swipe left on clickbait. combine w tip below (x’s new recsys generated a lot of slop it thinks you want) and i think timeline hygiene is the best its ever been - **dillon_mulroy** (2026-01-05): https://x.com/dillon_mulroy/status/2008039464925606226 - remember when we used to argue about tailwind - **thdxr** (2026-01-05): https://x.com/thdxr/status/2007967374822297782 - fyi i'm working on this as usual primitives first - you can bring whichever approach you want: worktrees, docker, cloud machines, whatever it's very hard, esp to do it without requiring a central server and have things work even if your laptop turns off https://t.co/pBwrt9SM0p - **thdxr** (2026-01-04): https://x.com/thdxr/status/2007960674404581507 - i'm working on the idea of every session running in a "workspace" workspaces can be local or remote (docker, cloud sandboxes like cf sandbox, your own vm whatever) and then your primary opencode session on laptop can be shut down and reopened and the remote workspaces will have continued - **levie** (2026-01-04): https://x.com/levie/status/2007958155137876183 - https://t.co/UVlRDGJu5I - **JustJake** (2026-01-04): https://x.com/JustJake/status/2007941551133982740 - https://t.co/V8siEv9fa5 - **lennysan** (2026-01-04): https://x.com/lennysan/status/2007892572740665853 - The best careers look like J-curves, not stairs "@Chamath drew me a picture on a whiteboard. He said the way a lot of people do careers is a set of stairs. You can be boring and stay on these stairs, just walk up the stairs and you'll get promoted every two years. But the much more fun careers are like jumping off cliffs. You jump off this thing, and you do fall for a period of time, but then this thing happens where you climb out. The picture he drew had this J-curve, that leads you to places that are way beyond where the stairs could ever get you. And to be totally honest that, has been my experience. Taking risks, accepting the terrible fall, and that experience of falling has been more than worth it." — @molly_g - **dillon_mulroy** (2026-01-04): https://x.com/dillon_mulroy/status/2007874356857442383 - we're live! deep dive on implementing DOOM in typescript types! https://t.co/qbWVLeT0tw - **elithrar** (2026-01-04): https://x.com/elithrar/status/2007849937342754877 - happy birthday to @ritakozlov! 🎉 https://t.co/2zOOsAe45C - **threepointone** (2026-01-04): https://x.com/threepointone/status/2007842239012126805 - (written as a note to myself, sharing here.) for those entering 2026, feeling left behind/fomo/overwhelmed by the rate of change: 1. don't. it's so incredible early. social media lies to you. 2. establish a session of practice for every day. start with 15 mins, where you _force_ yourself to try a new tool/thing/ai on something you're already familiar with, purposely to not be biased by your old workflow(s), and see if you can articulate what's good/bad, what could be better about the new thing. do this honestly. repeat. 3. expand the sessions. ask other people if they can show you how they use these things (like, literally on a call with screen sharing) 4. be mindful. david gilmour isn't the best because he shreds, but because he makes every note count. 5. be healthy. it's a slog from here on out. godspeed. - **techgirl1908** (2026-01-04): https://x.com/techgirl1908/status/2007794071340024109 - @flybayer try RPI. this process should work with any agent https://t.co/kPjetWXBqe - **thdxr** (2026-01-04): https://x.com/thdxr/status/2007648045526401038 - we finally shipped thinking in opencode - press ctrl+t to cycle through various thinking levels it's weird this came so late but it's a good example of how we ensure opencode stays extremely customizable while still working well out of the box quick video breakdown https://t.co/V0pRYYHZbn - **martin_casado** (2026-01-04): https://x.com/martin_casado/status/2007622894327869755 - Software is no longer the bottleneck. - **thdxr** (2026-01-04): https://x.com/thdxr/status/2007607067608265149 - it's so easy to make an opinionated tool of course you can make a great experience for someone while eliminating 99% of the market it's way harder to figure out how not to have an opinion - **ctatedev** (2026-01-03): https://x.com/ctatedev/status/2007579593960747316 - Testing a wild experiment today https://t.co/7TepPRthMP - **virattt** (2026-01-03): https://x.com/virattt/status/2007540767418687771 - @karrisaarinen Middle-out compression - **karrisaarinen** (2026-01-03): https://x.com/karrisaarinen/status/2007534281011155419 - https://t.co/RU33K24RsJ - **BraydenWilmoth** (2026-01-03): https://x.com/BraydenWilmoth/status/2007530341578514688 - We're getting closer to matching that Windows 98 vibes AIM messenger experience. 😎 https://t.co/1C9vnAR5Zb - **threepointone** (2026-01-03): https://x.com/threepointone/status/2007529930624532753 - I’ve been thinking about what the actual “day to day” looks like, particularly for the younger folks, and I feel like I’m directionally correct here. Corollary: bullish on systems that let you operate in this way. Linear comes to mind. https://t.co/8ahrGHDgif - **elithrar** (2026-01-03): https://x.com/elithrar/status/2007510017230991452 - built an @opencode git worktree plugin & tool by: 1. asking opencode to plan it in a GitHub issue 2. having it stand up a PR 3. reviewing itself by using two concurrent worktrees code: https://t.co/Hz5Xcrcf6g https://t.co/7UrVLsUpRp - **DanielleMorrill** (2026-01-03): https://x.com/DanielleMorrill/status/2007508036584341899 - executives, managers of people, managers of process, managers of agents… get Claude Code to be your chief of staff by organizing a repo around everything you do and find ways to make it better and make it clearly documented https://t.co/Wk9gabmE6I - **threepointone** (2026-01-03): https://x.com/threepointone/status/2007497491294396565 - new post: where good ideas come from (for coding agents) https://t.co/CwE393F8rz thought about why some people cook with coding agents, and others bounce off. my take: llms are "thought completers" traversing idea-space - so good results come from good navigation: constraints, context, oracles, loops. I used steven johnson’s "where good ideas come" from as a rubric and mapped it onto coding agents. also a note about how "juniors" can adapt to this new feature, and what "creating software" looks like practically. - **thdxr** (2026-01-03): https://x.com/thdxr/status/2007343974969942058 - before if you rejected a subagent's permission request it'd just get stuck now you can steer it in another direction https://t.co/68VFVPVi2E - **SherryYanJiang** (2026-01-03): https://x.com/SherryYanJiang/status/2007255902227968295 - one underrated thing i’ve been doing w/ opus 4.5 in cursor: i ask it to sketch the first design as terminal charts. super fast way to see structure + flows before i touch real ui. feels like thinking in ascii rn. https://t.co/1Hjd3sFeUI - **emollick** (2026-01-03): https://x.com/emollick/status/2007249835465072857 - When you see how people use Claude Code/Codex/etc it becomes clear that managing agents is really a management problem Can you specify goals? Can you provide context? Can you divide up tasks? Can you give feedback? These are teachable skills. Also UIs need to support management - **colinhacks** (2026-01-02): https://x.com/colinhacks/status/2007239546866667843 - git worktree is an incredible primitive with comically terrible DX - **fayazara** (2026-01-02): https://x.com/fayazara/status/2007129864965603554 - Cloudflare Tip of the Day Durable Objects have something called alarms. They’re like setTimeout() that just don’t die. Even if your Worker restarts, scales to zero, or gets redeployed - the alarm still fires. https://t.co/wVyBFdxQDp - **LukeParkerDev** (2026-01-02): https://x.com/LukeParkerDev/status/2006976943384866849 - I added full OTEL to @opencode There is no UI that comes close to this. Is this the year of @aspiredotdev ? OpenCode performance issues will be easy to view here, especially with metrics, like memory usage that you can correlate with spans cc: @davidfowl https://t.co/pHtB3Kvkre - **dillon_mulroy** (2026-01-02): https://x.com/dillon_mulroy/status/2006967589160001765 - this is fascinating https://t.co/BG24d9ohsN - **mattrothenberg** (2026-01-01): https://x.com/mattrothenberg/status/2006758554708164957 - You can build pretty much anything with opentui/react. Here's a basic React Flow inspired node-based UI for pipelining @replicate models with panning and inline editing, along with another generative thingy https://t.co/8juTFMPje8 - **dok2001** (2026-01-01): https://x.com/dok2001/status/2006736670499233830 - In 2026, taste is the engineering differentiator. Building is easy now. Knowing what to build, and what not to, is the hard part. Listen to users. Be elegant. - **matt_slotnick** (2025-12-31): https://x.com/matt_slotnick/status/2006460322354442320 - didn't see this coming... 2026 is going to be weird https://t.co/FkdqKsIJUf - **thdxr** (2025-12-31): https://x.com/thdxr/status/2006443405988319472 - 1 month later 650,000 monthly actives and it was a holiday month https://t.co/6NZPyfGHWp - **whoiskatrin** (2025-12-31): https://x.com/whoiskatrin/status/2006392754494898332 - Pretty good year if you ask me. The team grew a ton, just sent the last offer of 2025. We released updates to Agents SDK and Sandbox SDK every few days on average, slowly stacking MCP updates, improving transports, and sandboxing with code interpreters, @opencode integrations, Git, streaming, and a lot more. We focused hard on the foundations for agents. Agents need computers, and building both is a journey. Building these blocks has been one of the most satisfying things I’ve worked on. Grateful every day for the people beside me: @threepointone @ghostwriternr @mattzcarey @caise_p Ready for 2026! - **sdand** (2025-12-31): https://x.com/sdand/status/2006371094555664766 - introducing vmux: `vmux run` bundles and runs your code in a long-running cloudflare container and gives you a preview url in under 5s its a drop-in for uv run - close your laptop, run a tinker training job, come back and reattach via tmux! https://t.co/h6hasAj2ZI - **alex_verem** (2025-12-31): https://x.com/alex_verem/status/2006304233486995760 - What actually works: Hierarchical retrieval with compression. Instead of flat embedding space, build a tree structure with progressive summarization. Think: Encyclopedia → Chapter → Section → Paragraph Reduces search space from 50k to ~200 at each hop. https://t.co/olhsjYPetU - **matt_slotnick** (2025-12-31): https://x.com/matt_slotnick/status/2006157186062508424 - interesting discussion but i think it's tap dancing around the point to some degree. the question isn't whether you do chatgpt vs SaaS, and it's not about the GUI the future is about the capabilities of agents, whether you're OpenAI or SaaS. the method of interaction with the agents is an implementation detail you will care about the work that agents can do, you will not care about your interaction method with the agent in the same way you don't care when delegating work to a person whether you make the request by email, slack, over the phone, or wherever else. it's about the work being done. they either do the work or they don't everyone is in the business of selling agents that do work, whether they realize it or not. and those agents will be interacted with from an increasing number of places as we move into the future - **dillon_mulroy** (2025-12-30): https://x.com/dillon_mulroy/status/2006145400634298795 - @BenjDicken https://t.co/HqxqOu9vL3 - **BenjDicken** (2025-12-30): https://x.com/BenjDicken/status/2006114799633985823 - @dillon_mulroy https://t.co/am3HvQMvr9 - **dillon_mulroy** (2025-12-30): https://x.com/dillon_mulroy/status/2006106689032839326 - i wanted a better "human in the loop" flow for opencode such that i could easily make edits to proposed changes before/as they're suggested. so i made it https://t.co/OrP0jRjnM4 - **ZhangWeiHong9** (2025-12-30): https://x.com/ZhangWeiHong9/status/2006075321821376881 - 🚀 Single-agent coding assistants hit a wall on long-horizon tasks. Even frontier AI IDEs like Claude Code rely on manual sub-agent design—which is tedious and suboptimal. What if we could discover the optimal agent hierarchy automatically? Introducing BOAD: Bandit Optimization for Agent Design. 🔥 Outperforms GPT-4 & Claude-3.7-Sonnet (36B model) 🤖 Uses Multi-Armed Bandits for principled agent discovery Stop guessing prompts. Start optimizing. 🧵👇 This work is led by @iris_j_xu during her internship at MIT-IBM Watson AI Lab. Paper: https://t.co/U576P0979m Code: https://t.co/R2FfcBoSKu #LLM #RL #AIagents #CodingAgents #MultiArmedBandit - **ritakozlov** (2025-12-30): https://x.com/ritakozlov/status/2006070377005347101 - you should follow @michellechen https://t.co/dZgcIzVdoV - **michellechen** (2025-12-30): https://x.com/michellechen/status/2006056769693573534 - learning machines: model compilation and intermediate representations follow along as i build my mental model of how inference is optimized at the ML compiler level! today is a general overview of the stack, tomorrow i’ll go into example optimizations like kernel fusion - **elithrar** (2025-12-30): https://x.com/elithrar/status/2006028034889887973 - hacked together a quick web-perf agent skill for assessing FCP/LCP/CLS using the chrome-devtools MCP server. being able to run it from within a repo == the output becomes context to make immediate improvements. https://t.co/QqZqCVFAZD https://t.co/pqd3yuRahU - **davidlee** (2025-12-30): https://x.com/davidlee/status/2006008258192978156 - If my memory is correct Benchmark wasn't "ridiculed" for investing in Manus, they were chastised. Which is worse. Xenophobic, binary thinking. Reminded me of the SV chatter when the Russian mobster Yuri Milner invested in Facebook in 2009 - **boristane** (2025-12-30): https://x.com/boristane/status/2006000823893979236 - more spots to join my team we're hiring for workers observability 😁 https://t.co/KWW62l5vG3 - **zebassembly** (2025-12-30): https://x.com/zebassembly/status/2005988231326884175 - If you like observability, getting to build beautiful frontends and getting your hands dirty in distributed systems at scale, then please apply! I've been on this team for over a year now and it's been my favorite team since joining Cloudflare. https://t.co/CtimPhFhVD - **ilyasu** (2025-12-30): https://x.com/ilyasu/status/2005865829745795407 - interesting https://t.co/MtolaxhNrd - **zhengdongwang** (2025-12-30): https://x.com/zhengdongwang/status/2005848098531106916 - In my annual letter this year, I really try to get at what it means to "feel the AGI." Featuring compute, inevitability, second-order effects, travel tips, Andor, and Isaiah Berlin. https://t.co/tnamR3EWL0 - **brian_lovin** (2025-12-30): https://x.com/brian_lovin/status/2005816823887876209 - This blog is aimed at software engineers, but I think most posts apply to design as well. If you're a designer, I recommend digging in: https://t.co/IiE4e8zfu8 A few good recent posts: • You can't design software you don't work on https://t.co/DgmhizJpfV • We are in the "gentleman scientist" era of AI research https://t.co/yUb0oSDFme • Seeing like a software company https://t.co/ohOxOTjgxH - **boristane** (2025-12-29): https://x.com/boristane/status/2005737695104889004 - ai software engineers, ai code reviewers, ai sres, etc. we're at the cosplay stage of a fundamental change in this industry every process must change - **dflieb** (2025-12-29): https://x.com/dflieb/status/2005674455474000324 - Hit Christmas gift this year: Stickerbox. Press button, describe something, AI generates an image and prints on a sticker. Great product design. We’ve printed hundreds of stickers so far. Made by one of my YC Summer 2009 batchmates! https://t.co/BjJvvKh71a - **thsottiaux** (2025-12-28): https://x.com/thsottiaux/status/2005157101577142623 - Background Terminal is a Codex feature we’ve been baking for a bit and will soon turn on by default. More powerful than it sounds, it allows to Codex to multiplex work and significantly speeds things up. Default-on within OpenAI. You can enable with /experimental in the CLI. - **bznotes** (2025-12-28): https://x.com/bznotes/status/2005100249358963161 - (Almost) nobody enjoys paying taxes. Especially as you get richer you have more friends who do all sorts of circus tricks to avoid paying taxes. And you feel like you should be doing some of that too. Rich people end up privatizing much of their life. And then they don’t like why others depend on government subsidies to make ends meet. This seems to be a universal theme. - **martin_casado** (2025-12-28): https://x.com/martin_casado/status/2005081959576600709 - @RoKhanna Antagonizing the people who made your district the amazing place it is, is a new low. The fact you don’t realize how tone deaf you’ve become is another indictment of your ability to represent moderates in your own district. - **karpathy** (2025-12-28): https://x.com/karpathy/status/2005067301511630926 - I was inspired by this so I wanted to see if Claude Code can get into my Lutron home automation system. - it found my Lutron controllers on the local wifi network - checked for open ports, connected, got some metadata and identified the devices and their firmware - searched the internet, found the pdf for my system - instructed me on what button to press to pair and get the certificates - it connected to the system and found all the home devices (lights, shades, HVAC temperature control, motion sensors etc.) - it turned on and off my kitchen lights to check that things are working (lol!) I am now vibe coding the home automation master command center, the potential is 🔥.And I'm throwing away the crappy, janky, slow Lutron iOS app I've been using so far. Insanely fun :D :D - **mattsimpsn** (2025-12-27): https://x.com/mattsimpsn/status/2005057576904655060 - react-devtools → terminal coming soon in @opentui/react https://t.co/QDpkVzmhbT - **johnarnold** (2025-12-27): https://x.com/johnarnold/status/2005046740622454795 - The broader trend is that American manufacturers are gaining market share at home while steadily losing it overseas. https://t.co/Xeg6aDKh5j - **sdamico** (2025-12-27): https://x.com/sdamico/status/2005024588049297787 - if California needs to raise taxes, there's an approach that is extremely equitable. A ~$10K parcel tax with a refund that approaches 100% if you bought the home recently, and 0% if you bought it before 1976. Believe this can be passed locally without reforming prop 13. - **sdamico** (2025-12-27): https://x.com/sdamico/status/2005022424065351966 - the open secret: the goal is to depopulate places like LA and SF of tech instead of having to cross what is seen as a "bridge too far": build housing in expensive areas and reform prop 13. https://t.co/RC0meJzNYR - **techgirl1908** (2025-12-27): https://x.com/techgirl1908/status/2004972521087541463 - We desperately need a new approach for reviewing PRs. Modern software engineering is resulting in a surplus of PRs and the tools available aren't quite hitting the mark. It's not just that the code needs reviewing, the queue needs managing, the PRs themselves need to be smaller and stacked. Graphite seems closest to solving this... but yet still something is missing. Looking forward to seeing more innovation in this space in 2026. - **mnolangray** (2025-12-27): https://x.com/mnolangray/status/2004969619866718687 - Spending a lot of time in Central America, as I have done every Chrisrmas for the past decade, and I think Americans have been shielded from declining consumer-facing soft power, especially vis a vis China. - **thdxr** (2025-12-26): https://x.com/thdxr/status/2004661882607534080 - our team #1 bottleneck is finding clarity figuring out what we should do, how we should do it, what's the design, how it's positioned, how it'll evolve this is a fortunate position to be in but this is why we aren't 100x more productive with ai - **pdrmnvd** (2025-12-26): https://x.com/pdrmnvd/status/2004609154850464025 - oh you’re using claude code? everyone’s using open code. just kidding we’re all on amp code. we’re using cline, we’re using roo code. we just forked our own version of roo. were using kilo code. we were on coderabbit but their ceo yelled at us so now we’re using qorbit. apple just acquired them for $30bn so we just migrated our entire team to slash commands. one guy is still on aider. the PM is on loveable. he just shipped a new product on replit. the intern installed a slackbot that lets you chat with your spreadsheet. legal is still reviewing devin’s enterprise contract. we evaluated junie for three ukrainians using jetbrains. someone in slack just asked “has anyone tried amp?” we are using goose for scripts. next week we’re piloting augment code. the CTO heard good things about trae.​​​​​​​​​​​​​​​​ our CEO is friends with the guy from conductor. our CFO resigned. our CISO said we’ve had fourteen supply chain attacks in the last week. we’re shipping the worlds most expensive todo app. - **karpathy** (2025-12-26): https://x.com/karpathy/status/2004607146781278521 - I've never felt this much behind as a programmer. The profession is being dramatically refactored as the bits contributed by the programmer are increasingly sparse and between. I have a sense that I could be 10X more powerful if I just properly string together what has become available over the last ~year and a failure to claim the boost feels decidedly like skill issue. There's a new programmable layer of abstraction to master (in addition to the usual layers below) involving agents, subagents, their prompts, contexts, memory, modes, permissions, tools, plugins, skills, hooks, MCP, LSP, slash commands, workflows, IDE integrations, and a need to build an all-encompassing mental model for strengths and pitfalls of fundamentally stochastic, fallible, unintelligible and changing entities suddenly intermingled with what used to be good old fashioned engineering. Clearly some powerful alien tool was handed around except it comes with no manual and everyone has to figure out how to hold it and operate it, while the resulting magnitude 9 earthquake is rocking the profession. Roll up your sleeves to not fall behind. - **zarazhangrui** (2025-12-26): https://x.com/zarazhangrui/status/2004554163125141699 - https://t.co/QfJPQXi1cO - **TimMLatimer** (2025-12-24): https://x.com/TimMLatimer/status/2003893007758901564 - Hardtech development cycles are so long you almost have to start a company at a time when your sector is out of favor and hope it comes together once the tech is ready to get the timing right. If your sector is hot and you’re just raising a seed, probably NGMI. - **Founder_Mode_** (2025-12-24): https://x.com/Founder_Mode_/status/2003867132069630074 - Jack Dorsey on becoming a better storyteller: "I found myself very early on thinking about something like thinking about this early idea for Twitter and saying to myself, I could build this awesome. You have those shower-like moments, or you're walking at midnight in some town in New York City, and you've got these amazing brand ideas. And then you start thinking, well, I could really start doing this if only X and if I had this person or if this technology existed or if this happened or this happened. And what I realized was that I was constantly making excuses for not working on it. And then the window had passed, and then I couldn't do anything. So I think it's really, really important to write it out or to draw it out or to code it. But you need to get it out of your head. And the reason you have to get it out of your head is that you need to be able to see it on a surface that is not in your mind. And once you can see it, and once you can step back from it, then you can also decide this passes my filter, my constraints, so maybe I can show it and share it with some other people. And then they will be like that's the stupidest idea ever and or that's somewhat interesting, but maybe this and this and this. So the sooner you can do that, then you have a lot of momentum around it, and you can really decide if you want to commit to it and work on it more or put it on the shelf for a later date. And the realization that I think everyone needs to have about that latter option, putting it on the shelf, is that you can come back to it and it will surface back up in another piece of work or another idea at some point in your life. So having that ability to close off a chapter and move on is really, really important. You can't have all these open threads, and that's what I realized I was doing. And that also encouraged me to really write more and to really think about what's the story? How are people coming to this? And like when I show my friends this, how are they going to react and I would write it down. I would actually treat it like a play. And when I realized that I was writing plays, I read a lot more plays for style and for substance and for technique and I think it's really good. I think there is another company that I have always looked towards for inspiration and I know a number of people in this room probably have a similar company in mind, which is Apple. Apple, I think, is run like a theater company. It has a great sense of pacing, has a great sense of story and has a great sense of execution and it's all about event-driven, it's all stage-driven, the stage being a billboard or the stage being a keynote or the stage being a product launch. All of it has a very, very cohesive end-to-end story. I mean you think about what happened when Steve Jobs came back to the company. The first thing he did was kill every product line the company was working on. And for two years,rs they had no product on the market whatsoever. All they had were a bunch of posters all around the world with Steve Jobs' heroes, and it said, think different. And it was just focused on bringing up the brand and making people aware of the brand again and how the brand is aligning to this particular feeling and story. And then they came out with the iMac and then built iTunes and then the iPod, and they realized that, wait a minute, people are carrying music on their phones now, so we better build a phone, an iPhone. And so this unfolding of the plot and the epic story has been very, very interesting to watch, especially if you look back to that time when he came back to the company. So I've learned a lot from that company and other companies that operate in a similar fashion." - **pingToven** (2025-12-23): https://x.com/pingToven/status/2003602037091762510 - re: what “OpenAI compatible” actually means in 2025 through the lens of gpt-oss reasoning_effort the term gets used a lot across the industry, and it carries way more implied guarantees than it should historically, “OpenAI compatible” has meant support for an OpenAI-style chat completions API. same high-level schema, same messages array, same basic parameters. this shape became the default largely due to OpenAI's first-mover advantage that API worked well enough to start. it was built quickly, for the moment, before tool calling, hybrid reasoning, structured outputs, or multimodal inputs were common. it's hard to fault the original design given the capabilities at the time but once you try to apply that same API shape across 60+ providers, inference engines, and model families, the cracks show up very fast. the practical result is that the same request can succeed, fail, or subtly change model behavior depending on where it runs. let's start with the messages array: on paper it's simple: ordered turns, each with a role and some content. in practice, this array is handled wildly differently. some providers support arrays of content types per turn which is typically used with text+image in a single turn, but could just be multiple text strings. others throw errors when you try to do this - others silently concatenate. some models were trained for it, others weren't, which means you may get degraded performance even when the request “works.” role ordering is another source of variance - for example, some providers accept a messages array with a single turn of system role, while others require at least one user turn. some allow assistant prefill and correctly continue generation - others allow prefill but with a specific parameter, and that parameter differs. others ignore it or throw errors. all of this can happen on the same model depending on where it's hosted that's before you touch sampling parameters even temperature ranges differ - some cap at 1, some allow higher. logprobs can come back in different shapes. newer OpenAI models don't allow modifying temperature or top-p at all, while open-source models still rely on them heavily. compatibility here often means 'best effort' structured outputs add another layer json object vs json schema, partial json schema support, streaming sometimes supported, sometimes not. some providers support reasoning plus structured outputs, others don't tool calling is where layers of variance really add up: tool calling is structured output plus special tokens plus a parser plus chat templating plus finish reasons. tool parsers are frequently incorrect, and that's not always the provider's fault - even when a model lab works with the popular engines like vllm and sglang, we see tool call parser issues well after launch. the kimi k2 vendor verifier project uncovered various problems in the inference engine implementations weeks after model launch tool_choice support varies by model and provider. auto usually works. none often breaks in subtle ways. forced is rare. function-by-name works on some stacks and not others. finish_reason=tool_call is not guaranteed even tool call IDs are inconsistent - regex expectations differ, length limits differ. reuse the same ID across providers and you will eventually hit a hard error somewhere. at this point, "compatible" describes the shape of the request, not the semantics reasoning_effort is a newer example of the same pattern, brought to the limelight by @xeophon's work with gpt-oss benchmarks. he surfaced a measurable variance caused by provider-level incompatibilities OpenAI introduced a new enum parameter, reasoning_effort. initially the enum was low, medium, high. then minimal, none, extra high. gpt-oss only supports a subset (low, medium, high). meanwhile, most open source reasoning models only support reasoning enabled or disabled (think GLM family, DeepSeek after v3.1). when gpt-oss released, none of the inference providers had support for the reasoning_effort parameter, as it was mostly used on OpenAI's proprietary models. everyone rushed to launch gpt-oss, and the parameter and its impact on the amount of reasoning was swept under the hood for months. eventually, @xeophon published a tweet of a benchmark he ran through OpenRouter showing a ton of providers not changing the amount of reasoning based on the effort value sent to the OpenRouter API. Xeophon originally blamed the providers, but as soon as I saw it I realized the issue was largely our fault - we hadn't implemented support for it for each provider. this was a miss on both the OpenRouter team and the providers given how new and underspecified the parameter was - to this day, most providers do not have the parameter or supported values documented anywhere, and most providers did not communicate with us when support was added. we often have to chase teams down to tell us about their apis, as we need a deep understanding of implementations to be able to properly support user intent being transformed to upstream acceptable values. the fix for this could have been that we added piecemeal support for each provider - but i wanted to avoid this problem entirely in the future, so instead i spent a few weeks refactoring a ton of code to implement model-level reasoning configs. this means that in our database i can now specify a few things: - whether a model supports reasoning effort - what values of the enum it supports (out of none/minimal/low/medium/high/xhigh) - and what the default value should be this solves multiple issues for us in a scalable way: - it prevents upstream APIs from throwing errors if we were to pass reasoning_effort param to models that don't support it - it prevents upstream APIs from throwing errors if we were to pass a value that is not supported - and it normalizes the default effort value across providers to ensure that if the user doesn't specify, the behavior is consistent. once this model level config implementation was done, we still needed to ensure that we were plumbing the values into the right fields for different providers that expected it in different places. some expect it in chat template kwargs, others in a top-level reasoning object. once all this was done, @xeophon was able to get consistent reasoning effort behavior across all providers (minor caveat for bedrock API, which we fixed after communicating with their team about their unique param implementation.) so now, for reasoning_effort specifically, the OpenRouter experience should be much better. and in the future, any new models with effort support will be much easier for us to support, and we will be able to move more quickly with them. and all of that's just the request side. here’s a quick, non-exhaustive sample: token accounting differs - counts for cached tokens, reasoning tokens, image tokens etc differ. sometimes reasoning is returned separately. sometimes wrapped in tags. finish reasons are not consistent even across proprietary APIs. what we do at OpenRouter all day is deal with this reality. we benchmark constantly and try hard not to change model behavior. and we recently added debug tooling so you can see the exact upstream request if something looks wrong (check that out here: https://t.co/ckrcOESJZL) special thanks to @xeophon for digging into this, working super hard on it, and helping our team fix the issues on our API and even help upstream providers realize how important it is to document these nuances. the takeaway isn't that anyone is doing things badly. this space is just evolving faster than the original API shape ever anticipated. most of this complexity is invisible when things work, and painfully obvious when they don't. if you find this stuff interesting, or if you enjoy untangling invisible incompatibilities that break real systems, we're hiring! - **hwchase17** (2025-12-23): https://x.com/hwchase17/status/2003599022871777467 - https://t.co/s3aNKxiJV0 + skills - both open standards, and can combined with an agent harness (claude code, codex, deepagents) can pretty much define a custom agent currently there's no way to package those together - claude plugins is close but i think lets you add an https://t.co/s3aNKxiJV0 am i alone in wanting this? or do other people want as well? - **tobi** (2025-12-23): https://x.com/tobi/status/2003586628145807568 - I'm really really excited about Agentic Commerce. There is so much amazing stuff being built. Everything I test just feels delightful and right. Shopify is building all the layers of infrastructure to power a new cambrian explosion of creativity in shopping - **JayaGup10** (2025-12-23): https://x.com/JayaGup10/status/2003525933534179480 - https://t.co/uPXcTUEsnc - **jayparsons** (2025-12-23): https://x.com/jayparsons/status/2003525774759068004 - Also, everyone should be aware that "luxury" is a marketing term only. Usually nowhere near what you probably consider "luxury." It's just another term for market rate apartments that actually pencil out. - **jayparsons** (2025-12-23): https://x.com/jayparsons/status/2003504702319919426 - Too many people say "we don't need more luxury apartments." If you care about affordability, yes you do. Great article from Bloomberg that shows when you build "luxury" apartments, even lower-income renters benefit from the downstream impact. Science > Rhetoric https://t.co/7csiPqEN5O - **theo** (2025-12-23): https://x.com/theo/status/2003361165032988721 - Okay I knew OpenCode was improving fast but....damn. I didn't recognize their game. https://t.co/lUFFPrBdUr - **ghostwriternr** (2025-12-22): https://x.com/ghostwriternr/status/2003142227506213336 - Nerd-sniped myself: Remote MCP running OpenCode on Sandboxes! 🧑‍🎤 - Kickoff tasks from ANY client - List & resume sessions - Auto backups to pickup old sessions anytime - Proxy layer to authenticate requests from sandbox - (Optional) View session via web UI Happy hacking and holidays!🎄 - **mattdorsey** (2025-12-21): https://x.com/mattdorsey/status/2002865135942242422 - But the problem ISN’T that the utility is a monopoly — most utilities are — it’s that as an INVESTOR-owned utility, @PGE4Me prioritizes shareholders over ratepayers. That’s a feature, not a bug. And it’s why @SFGOV’s municipalization bid deserves far more news coverage than it’s getting. Here’s the latest from that ongoing @californiapuc proceeding… (2/3) https://t.co/loOSNKOmWp - **DrSamuelBHume** (2025-12-21): https://x.com/DrSamuelBHume/status/2002826538140803135 - 10 breakthroughs in medicine in 2025 Read it here: https://t.co/YyaYTKd3Cs https://t.co/mNgirpMtRu - **techgirl1908** (2025-12-21): https://x.com/techgirl1908/status/2002764463431430345 - "Skills just killed MCP" is one of those tell-me-you-dont-understand-agentic-stacks moments. my latest blog post helps clarify the purpose of each https://t.co/0sLr5zG7kM - **grinich** (2025-12-20): https://x.com/grinich/status/2002275293734609176 - 28% of these are powered by @WorkOS 📈 https://t.co/35pHQUM5c4 - **zeeg** (2025-12-19): https://x.com/zeeg/status/2002166002214404190 - Imaging building an error prediction tool and knowing you're sitting on the dataset that contains basically every bug known to mankind 🤓 - **gooby_esq** (2025-12-19): https://x.com/gooby_esq/status/2002158726153707925 - Anyone using GEPA to evolve clause https://t.co/pLDartjmea files yet? - **thdxr** (2025-12-19): https://x.com/thdxr/status/2002048783341731946 - been reworking our permissions system last few nights it's mostly a design problem - i had a good experience using opencode to describe all the "user stories" i pointed out ones it missed then we laid out all the different options and which stories they didn't handle well helped me get a full understanding of the problem - then i laid it out on a call with aiden he pointed out one constraint i hadn't considered dropping and when we removed that a much better solution showed up then i did the implementation by prompting opencode this flow is what the "ai will do everything" people won't tell you about - **whoiskatrin** (2025-12-19): https://x.com/whoiskatrin/status/2001990900025627113 - We put together a personal AI agent for discord that feels stateful by default. It remembers context and never runs out of it, uses tools through MCP, and comes with a dashboard. Built with Agents SDK All the kudos go to @caise_p for putting it together https://t.co/qQRdEgVxxp - **mitchellh** (2025-12-19): https://x.com/mitchellh/status/2001810354096214059 - The lack of "feature design" is why so many products over time feel hollow or messy. This isn't visual design. This isn't architectural design. I thought that a short video lecture of what feature design is and a real case study of applying it in Ghostty would be helpful. Feature design is the planning step behind how you're going to solve one or more user problems with a product feature: what that feature looks like, how it feels, and not just how its going to tactically solve these specific problems, but how that solution is going to interface with the edges of other features that currently exist or are planned to exist in the future. - **threepointone** (2025-12-19): https://x.com/threepointone/status/2001804871733830042 - duuuuuurable objects https://t.co/NuPz7UIFe8 - **KentonVarda** (2025-12-18): https://x.com/KentonVarda/status/2001803932495192537 - It's true, I had the idea and the first thing I did was write the blog post assuming it worked. That got people interested and, luckily, @threepointone then proved it did work. And that's how you make things happen at Cloudflare. 😅 Nobody tried to come up with a better name. https://t.co/TYxS1wa0jD - **dok2001** (2025-12-18): https://x.com/dok2001/status/2001800004336574691 - It's great to see everyone adopting Code Mode! Like most good things @Cloudflare, it started with a blog post by @KentonVarda. He wrote it before the code, and we never got around to figuring out a better name. https://t.co/jNi2OOiLPF https://t.co/czR5cO4Ec8 - **floydophone** (2025-12-18): https://x.com/floydophone/status/2001783685100536041 - @swyxio I go back and forth on this so often and I’m not sure where I stand currently. However, what I have observed is that ~100% of the time that engineering asks for more PMs it’s not actually about PM bandwidth - **yomnashousha** (2025-12-18): https://x.com/yomnashousha/status/2001758726978666687 - Deploying static sites to Workers has never been easier!! Just import your repo in the dashboard and set npx wrangler deploy --x-autoconfig as your deploy command 🚀🚀🚀 Watch the video to see how (+ includes sneak peak into what's coming next) ✨ - **grinich** (2025-12-18): https://x.com/grinich/status/2001736197576757541 - @signulll every ARR announcement should be required to also include margin - **mwylde** (2025-12-18): https://x.com/mwylde/status/2001733763903144136 - 1/ Hot on the heals of R2 SQL aggregations, we have one more release from the @CloudflareDev Data Platform to close out the year: *Iceberg snapshot expiration*. So what is snapshot expiration and why does this matter? - **ankrgyl** (2025-12-18): https://x.com/ankrgyl/status/2001727158847443012 - I've been fortunate to work on the bleeding edge of data systems for over a decade now. But nothing compares to what we're seeing with agents. People want to: * Ingest 20gb+ traces * View them immediately without ingest delay * Search through billions of them to debug specific issues or find trends What gets me excited is that when we first rolled out Brainstore just 8 months ago, ALL of these requirements were easier. I can't wait to see what our customers push us to build over the year to come. - **MarcSelwan** (2025-12-18): https://x.com/MarcSelwan/status/2001726992794956010 - It's been a fun week for the Cloudflare Data Platform team. We've made some big improvements: - Aggregation support in R2 SQL: https://t.co/9Mhl94V9Hv - Snapshot Expiration in R2 Data Catalog: https://t.co/HQoy6kaZQr These go hand in hand because the metadata cleanup/management that snapshot expiration helps with will speed up performance of these aggregation queries, especially with compaction enabled. We've still got a lot to do and we're gearing up to start off the new year strong. - **burcs** (2025-12-18): https://x.com/burcs/status/2001721015689449967 - come help us shape the next 6 months... we are hiring. https://t.co/X46Ct5uJ2f - **ghostwriternr** (2025-12-18): https://x.com/ghostwriternr/status/2001719781154595095 - In the next version of Cloudflare Sandbox SDK 📦 An experimental transport option enables communication within the SDK to use websockets. So agents running on workers that access the sandbox as a compute environment can now perform any number of operations! h/t to @caise_p 🙌 https://t.co/YeIHA9o8tM - **elithrar** (2025-12-18): https://x.com/elithrar/status/2001710465974730956 - @basedanarki @PlanetScale Right in the Cloudflare dashboard, btw: https://t.co/PHK46ljn2X 😎 https://t.co/V3NkyZiFBg - **burcs** (2025-12-18): https://x.com/burcs/status/2001699274938290284 - add hidden="until-found" to let users cmd+f for content hidden in collapsed sections https://t.co/GM9LFI36EJ - **nandafyi** (2025-12-18): https://x.com/nandafyi/status/2001693448525492487 - 6 months ago to now https://t.co/hrMlry3gHV - **elithrar** (2025-12-18): https://x.com/elithrar/status/2001690510684213455 - workers for platforms ~= Functions as a Service as a Service. let your users deploy code that acts on your platform. your own types. control over egress and logging. instrument user code. provide your own SDK. make your platform programmable. https://t.co/6mCKweWH8F - **yenkel** (2025-12-18): https://x.com/yenkel/status/2001660766395474305 - will be interesting to see how the industry starts leveraging code mode and where it leads good to see @goose_oss leading the way https://t.co/MCKuFDbHEB - **goose_oss** (2025-12-18): https://x.com/goose_oss/status/2001659839282286929 - We just shipped code mode, so goose can generate and execute code against MCP tools instead of loading every schema into context. This means fewer tokens, better control, and more scalable workflows. https://t.co/S1U34hloKe - **aharvard** (2025-12-18): https://x.com/aharvard/status/2001651284026494987 - https://t.co/GfxIhqEPOw - **Geo_papic** (2025-12-18): https://x.com/Geo_papic/status/2001499594841374942 - This is my favorite macro chart of the decade. It shows - almost to the DAY - when one macro era ended and another began. The Chinese Leverage cycle was replaced by the US Profligacy cycle. Both were begun and ended by geomacro factors, not some macroeconomic fairy. https://t.co/9GpbjC7ey8 - **cjc** (2025-12-18): https://x.com/cjc/status/2001450602845585452 - Someone asked me what good backchanneling looks like. When @zoink interviewed me, he said: "I’ve talked to people you've worked with and heard you're intense." That opened up a real conversation about what they likely meant, when that intensity shows up, and how I think about it myself. Another example: when I was far along in the process to join Linear, @karrisaarinen backchanneled me with the Stripe founders. Not to replace his judgment, but to check whether his read on me matched theirs. That’s the difference. Backchannels should sharpen your view, not make the decision for you. - **aidenybai** (2025-12-18): https://x.com/aidenybai/status/2001447718267576817 - https://t.co/jkKBhBdALk - **elithrar** (2025-12-17): https://x.com/elithrar/status/2001402494514053579 - (by far) my favorite change is the new ⌘+K search the team landed. shortcuts for jumping into specific apps/databases/etc. good fuzzy matching. feels fast. been a delight to use. https://t.co/pdHP2QY3Iu - **burcs** (2025-12-17): https://x.com/burcs/status/2001393597250204088 - there's a bunch of other papercuts that got cleaned up as well, shout to @BraydenWilmoth @hkfoster @invisal89 @nandafyi and @RozenMD and the team for all their hard work this is only the beginning, a lot more exciting stuff coming to a cloudflare dashboard near you - **burcs** (2025-12-17): https://x.com/burcs/status/2001393591852109863 - we recently shipped a number of quality life improvements for navigating cloudflare... we improved sidebar and top navigation making it easier and cleaner than ever to understand your current context https://t.co/B9Ew2aww4J - **burcs** (2025-12-17): https://x.com/burcs/status/2001393593819250973 - also added a brand new search/command+k experience to make navigating the dashboard frictionless you can finally filter by resource types and search for resources by their names https://t.co/tR1U40fo6Z - **burcs** (2025-12-17): https://x.com/burcs/status/2001393595564032409 - we even made it easier than ever to navigate in and out of zones and into the rest of the cloudflare platform (i needed this one so bad, i always got lost on this and i work here) https://t.co/lCwPyM6fSr - **cjc** (2025-12-17): https://x.com/cjc/status/2001372465361228222 - If you’re overreliant on backchanneling exec candidates, you’re not doing diligence. You’re outsourcing judgment. I wrote more about how backchanneling is becoming a crutch: https://t.co/uWoiuCV2r4 - **zeeg** (2025-12-17): https://x.com/zeeg/status/2001344236399280142 - Someone woke up yesterday, decided plan mode wasnt important, and then a bunch of people on the internet decided to broadcast it in the echo chamber. Stop trusting these people. They are not experts. - **southpolesteve** (2025-12-17): https://x.com/southpolesteve/status/2001318955646882082 - https://t.co/PnEQpxHXjV https://t.co/5AVt38RWvk - **glangley** (2025-12-17): https://x.com/glangley/status/2001127450500018663 - Oakland debated, and the facts won out. • Fact — after Flock tech was deployed in Oakland in 2024, violent crimes were solved at an increase of +11%. • Fact — Flock tech is instrumental in solving 1 in 10 crimes in Oakland today. • Fact — a coalition including the @OaklandNAACP and Oakland Chinatown Chamber of Commerce made their voices heard tonight in favor of the Flock contract. And finally, fact — tonight, the City Council voted 7–1 to continue using the tools proven to reduce crime and protect the community. - **zebassembly** (2025-12-16): https://x.com/zebassembly/status/2000962198118518931 - Workers has (rightfully) earned a reputation that on-boarding your existing app can be difficult, and we're looking to change that! Now in Wrangler we have an experimental flag to enable "auto-config" where we'll automatically cloudflare-ize your app. Run `npx wrangler deploy -x-autoconfig` to deploy to Region: Earth - **elithrar** (2025-12-16): https://x.com/elithrar/status/2000957744162230458 - more "it should just work" things coming from the Workers team! 👇 https://t.co/UlXn7vD7IH - **MattieTK** (2025-12-16): https://x.com/MattieTK/status/2000951042863980792 - 🆕 Today in wrangler 4.55, experimental automatic configuration for your web framework for Cloudflare Workers. npx wrangler deploy --x-autoconfig We support Next.js, Astro, Nuxt, and many more, + static sites. Watch the demo below and read more in the link in the reply. https://t.co/NyaYAhkbKd - **ernietedeschi** (2025-12-16): https://x.com/ernietedeschi/status/2000742239274742157 - The US health care sector actually has *below-average-level* productivity for the US (~1/3 lower than the avg), so its higher US weight is mechanically *lowering* average US productivity levels rather than raising them (before even getting to sectoral substitution considerations) https://t.co/R0UTPp2wvi https://t.co/agGKboBGEP - **Vjeux** (2025-12-16): https://x.com/Vjeux/status/2000732151369883697 - @davidgu Could you bring that to puppeteer as well, they transmit screenshots from Chrome to the runner via websocket and it's a huge performance bottleneck in practice. https://t.co/nCWPIYhJhP - **matt_slotnick** (2025-12-15): https://x.com/matt_slotnick/status/2000715025481113637 - thought experiment: if cost to build software decreases, is there a world in which markets end up forming predominantly vertical-first and cross-functional? eg instead of sales software and support software, you have life sciences software that handles HR, IT, sales, support - **ritakozlov** (2025-12-15): https://x.com/ritakozlov/status/2000682880456880365 - last week i gave a demo of a code-mode agent and a good ol' MCP agent spoiler alert: the code-mode agent uses 80% fewer tokens than the vanilla agent 🔥🔥🔥 check out my full talk below for full talk thank you @workos for having me. fantastic event and such a great audience! https://t.co/1FBewGxiVg - **brian_lovin** (2025-12-15): https://x.com/brian_lovin/status/2000673685863358736 - I don't know how to say this in a smart way, but: I think constantly comparing software design to architecture is a mistake. Bits vs atoms. The physics are different. It's the go-to analogy for design debates, but software design and architecture feel like different enough practices that the comparison is adding more confusion than clarity. - **signulll** (2025-12-15): https://x.com/signulll/status/2000673417847333247 - most ppl completely misunderstand what storytelling means in this era. let me explain in very very simple terms. storytelling is not marketing (people are very confused by this), in fact it’s not really a marketing role at all. people think journalists can be story tellers (maybe, but not really). ppl think it’s creating ai slop, posting to x, creating launch videos, or brand theater. nope. this is pure garbage most of the time. real storytelling is a product & technology function. the real *why* behind anything you do or build. steve jobs was the one of the greatest story tellers ever. look at how he talks about a very simple new feature (the proximity sensor) on the original iphone. before iphone very few phones had it. look at crispness of he talks about a tiny thing… why does this exist? why should you care? then answers them in ~20 seconds flat. the explanation is emotional, practical, & instantly relatable. you feel smart for understanding it. you think, “oh of course.” this moment is storytelling. it compresses complexity into legibility. it turns engineering decisions into human meaning. that’s the storytelling craft. none of it is persuasion, & it’s certainly not hype. just pure sense making. this is why great founders are the greatest marketers. one explains what you should buy. the other explains why reality had to be this way. - **ajoshhoward** (2025-12-15): https://x.com/ajoshhoward/status/2000649885440455082 - I am hiring an engineer (at least one) to improve the Durable Objects developer experience. Tactically: - Maintaining @cloudflare/actors - Reducing Wrangler configuration - Managing our developer community, e.g. Discord Apply here: https://t.co/aBNP8USEEL - **dok2001** (2025-12-15): https://x.com/dok2001/status/2000634776055927062 - We are still seeing heavy activity from users probing for vulnerable React sites, with over 1 billion attempts in the first 11 days. Please upgrade React! https://t.co/amOlPLvLop - **terronk** (2025-12-15): https://x.com/terronk/status/2000626177770422616 - @JackieFielder_ Thanks for unblocking me. Glad it only took one email to your staff documenting your violation of my 1A rights. People are not displaced from their homes by commercial space being used for research. They are displaced by NIMBYs like you limiting housing supply. Hope that helps! - **eastdakota** (2025-12-15): https://x.com/eastdakota/status/2000622160159887476 - We’re seeing up to 14.5 million attempts per hour to exploit the #React2Shell vulnerability across @Cloudflare’s network. This is a very bad exploit. While our WAF is helping protect customers, it’s critical to update your React and Next.js instances as soon as possible. https://t.co/p1169aOYnC - **mwylde** (2025-12-15): https://x.com/mwylde/status/2000619563160133734 - The @CloudflareDev Data Platform team tried to keep this quiet by dumping the news on Friday afternoon. But this is too big to hide: R2 SQL, Cloudflare's distributed query engine, now supports aggregations! https://t.co/huw9JqieJn - **elithrar** (2025-12-15): https://x.com/elithrar/status/2000571303166681277 - so we just shipped an extensive best practices guide for Durable Objects. the goal: clear guidance + opinions on how to architect DO based applications, when not to use them, and how to use the storage, websocket & migration APIs. read it here: https://t.co/YaHNJp1EDn - **CloudflareDev** (2025-12-15): https://x.com/CloudflareDev/status/2000544243568271852 - Don't sleep on the Sandbox SDK. Create code sandboxes, checkout git repos, send them prompts and stream responses back directly from your application code: https://t.co/fCmE6ExhM4 https://t.co/X24RIZVJEv - **fayazara** (2025-12-15): https://x.com/fayazara/status/2000535913848623290 - Lots of improvements to flarechat - a customer support chat widget trained on your own docs - now using @CloudflareDev agents sdk - code syntax highlighting - has the conversation context I will open this up soon, once I add some guard rails using CF AI Gateway https://t.co/nOCymhtHuD https://t.co/EJd0gIuVLo - **MichaelArnaldi** (2025-12-15): https://x.com/MichaelArnaldi/status/2000480322497159184 - @dillon_mulroy If you don’t have a strong basis for code to be maintainable at scale Opus starts to fuck up in the same way that humans fuck up, to quote you “discipline doesn’t scale” applies to agents too. What changes is tolerance to verbosity and syntax, you won’t be writing the code soon - **clairevo** (2025-12-15): https://x.com/clairevo/status/2000407039965835536 - @GergelyOrosz Any engineer who can make good use of $35k in tokens is worth investing $35k in tokens. Someone great makes this look cheap. I think bigger problems is underutilization in Eng orgs. - **ankrgyl** (2025-12-15): https://x.com/ankrgyl/status/2000400463532114208 - Agents simplifying everything to file systems reminds me of Hadoop. Hadoop’s big idea was that analysts can just write scripts that access files directly instead of specialized interfaces like SQL. Substitute people with agents and now you’re in 2025… I think we will relearn the same lessons: modeling every problem as a file system is slow, brittle, and insecure at any interesting scale. The reason this happens every N years, however, is that as computing paradigms change, the abstractions we have built up to support them start to crack. File systems allow you to cut past them and reinvent things from scratch. In the data world, many of the ideas from Hadoop (eg parallel compute at scale) found their way into Snowflake and Databricks, which are much more sophisticated abstractions. I’m quite excited to see the new abstractions that emerge from our collective learnings right now. I’m also bearish that filesystems are the long term solution. - **grinich** (2025-12-15): https://x.com/grinich/status/2000398154542534864 - what does it mean? nobody knows.. but it's provocative! https://t.co/mtO23I3Akw - **dillon_mulroy** (2025-12-15): https://x.com/dillon_mulroy/status/2000392353761091855 - do we even? or do we still target the developer who is the driving force of the agents - who is considering the trade offs, the broader system, the blast radius of the decisions they’re making for the agent. in this way abstractions, libraries, frameworks, etc are a tool for communicating ideas - clearly - to the agents - **willahmed** (2025-12-14): https://x.com/willahmed/status/2000345992378654746 - ❤️🙏 https://t.co/2PZdyfoxhO - **simpsoka** (2025-12-14): https://x.com/simpsoka/status/2000338110899789841 - A continuous AI system: - Is triggered by environment, not user intent - Accumulates context over time - Needs memory, traceability, and control - Acts opportunistically, not conversationally - **nxthompson** (2025-12-14): https://x.com/nxthompson/status/2000332101187784975 - I’m getting old. But not too old to not love winning a slushy, snowy, somewhat insane 5k in Prospect Park. https://t.co/NE0mmApI7t - **dok2001** (2025-12-14): https://x.com/dok2001/status/2000313726118113405 - @JustJake Competition is good. - **thdxr** (2025-12-14): https://x.com/thdxr/status/2000247850337239530 - there are good programmers i followed before AI was a thing the way they use AI is pretty different than the people who are getting a following now specifically for how they use AI i know everyone is primed to think every tech shift flips the tables but i'm not betting on that - **ritakozlov** (2025-12-14): https://x.com/ritakozlov/status/2000240346052321789 - happy snow day ❄️ https://t.co/rd543MuNot https://t.co/rwwsubm5zB - **lennysan** (2025-12-14): https://x.com/lennysan/status/2000223601728029175 - "I ran a startup for 5 years. I thought we moved fast. Then I joined OpenAI—and realized I had no idea what fast actually meant." Head of Product for @OpenAI's Codex @embirico and I chat about: 🔸 How OpenAI built the Sora Android app (which hit #1) in just 18 days by using Codex 🔸 Codex growing 20x since August 🔸 How they're thinking about the dev bottleneck shifting from coding to reviewing AI-generated work 🔸 Why coding will be a core competency for every AI agent—because writing code is how agents use computers best 🔸 Why the bottleneck to AGI-level productivity isn’t model capability—it’s human typing speed 🔸 The vision of Codex, and how they plan to win Listen now 👇 • YouTube: https://t.co/HGGuxeutoA • Spotify: https://t.co/iG32qAHsdW • Apple: https://t.co/wCBi3I67CP Thank you to our wonderful sponsors for supporting the podcast: 🏆 @WorkOS — Modern identity platform for B2B SaaS, free up to 1 million MAUs: https://t.co/XH3bKmeEkH 🏆 @Fin_ai — The #1 AI agent for customer service: https://t.co/gHMbsr9U6W 🏆 @Jira Product Discovery — Confidence to build the right thing: https://t.co/QdYvsx2QyU - **swyx** (2025-12-14): https://x.com/swyx/status/2000130176915820881 - there’s a new energy in san francisco. if you weren’t here in 22-23 you might be forgiven for thinking this is normal. but it’s not. the parties are going past midnight. people are making Effort. everyone’s dating again, even the aces. people are drawn away to hard tech from b2b saas. no one, -no one-, is too rich or famous or important to meet somebody small who might have the next great idea. there’s even new free invite only startup driverless car services launching. running and working out are in, endless coffee chats and zirp zombies are out. nature is healing. san francisco is healing. i am healing. - **bslobodin** (2025-12-14): https://x.com/bslobodin/status/2000044130739552368 - @dok2001 turing tumble and/or spintronics i can vouch for personally, and bridge or bust looks great too. these folks don't miss. https://t.co/WVujCil82v - **ernietedeschi** (2025-12-13): https://x.com/ernietedeschi/status/1999942019859484679 - @justin_policy @CtheLala The apples-to-apples comparison then would be EU member states to US states. See here (the chart doesn't say so clearly but it is PPP-adjusted): https://t.co/lg9N7hLwHg https://t.co/LagDkXfZaw - **ernietedeschi** (2025-12-13): https://x.com/ernietedeschi/status/1999902198113992983 - Even if you adjust for hours worked, US real GDP per hour was 22% higher than the EU in 2023 and 15% higher than the Euro Area. https://t.co/5W2P2EJ6IT https://t.co/HhnPUWLSJS - **tokengobbler** (2025-12-13): https://x.com/tokengobbler/status/1999832968777416776 - I was sleeping on Cloudflare sandboxes, but now am awake - **GergelyOrosz** (2025-12-13): https://x.com/GergelyOrosz/status/1999809869541478543 - Gartner's business model crumbling: Anthropic and Claude are not on any Gartner list of AI coding assistants (Anthropic doesn't pay Gartner) and Cursor is downranked bad (they also don't pay Gartner) Yet large enterprises buy Cursor and Anthropic+Claude because they are GOOD https://t.co/A6ECKIIM10 - **M_C_Klein** (2025-12-12): https://x.com/M_C_Klein/status/1999622553057526201 - New at THE OVERSHOOT: The Fed Submits? https://t.co/O05KhOmMS6 America's central bankers have, in the aggregate, lowered the "appropriate" path for interest rates despite an improved growth outlook and a worsening inflation picture. There is no good explanation. https://t.co/sVljsAeBNt - **mscccc** (2025-12-12): https://x.com/mscccc/status/1999611820273729995 - As a person who worked on GitHub dot com and now builds PlanetScale using Rails + React. The answer to this question is: No https://t.co/oyrWUZvgbh - **garrytan** (2025-12-12): https://x.com/garrytan/status/1999596538960707895 - We are going to name and shame all the supervisors who voted for banning research and development in San Francisco LEGALIZE BUILDING TECH IN SF https://t.co/ykyWu9Y1of - **M_C_Klein** (2025-12-12): https://x.com/M_C_Klein/status/1999584195371250044 - @conorsen he's been saying he wants 1% since at least the summer, erdoganism is priced in - **terronk** (2025-12-12): https://x.com/terronk/status/1999532633097712090 - DSA Supervisor @JackieFielder_ is back on her bullshit. She wants to ban all R&D in the Mission, and it looks like she's going to get away with it bc the rest of the BoS isn't paying attention. Sorry this has to be a thread, but it's fucking insane and you need to know details. - **clairevo** (2025-12-12): https://x.com/clairevo/status/1999531464560435517 - The uncomfortable truth that’s going to be very apparent soon is that teams without PMs in the way (or w v unique “PMs” that know how to get out of the way) are going to be the fastest moving ones by a long shot. - **jaredpalmer** (2025-12-12): https://x.com/jaredpalmer/status/1999525369725215106 - We are sending out a proposal for Stacked Diffs on @GitHub to trusted design partners to gather initial feedback over the next few days. From there we’ll iterate and share the gameplan https://t.co/0CJubWV7gi https://t.co/ZUWjQCtmZJ - **leerob** (2025-12-12): https://x.com/leerob/status/1999513912652214525 - The cost of abstractions with AI is very high. Over abstraction was always annoying and a code smell but now there’s an easy solution: spend tokens. It was well worth the money to delete complexity from our codebase and it already paid for itself. - **corywilkerson** (2025-12-12): https://x.com/corywilkerson/status/1999487300703797351 - Innovator’s dilemma is brutally real. Tools I’ve loved for years are stuck (a trap you can’t really escape) while the unencumbered folks just crank on new things. It’s not even slow motion anymore, it’s happening so fast. - **ghostwriternr** (2025-12-12): https://x.com/ghostwriternr/status/1999448477424820714 - Out now in Sandbox SDK v0.6.5 📦 You can bring your own Dockerfile! We still publish 3 base images to use as-is or extend. But if you have an existing setup, adding the Sandbox binary is now only 3 lines. The binary will also take care of starting your custom scripts as well. https://t.co/Nw75H8XNBM - **MikeNomitch** (2025-12-12): https://x.com/MikeNomitch/status/1999335784995070086 - With all the recent upgrades to Python Workers, I thought I’d make some videos showing them off - going from a local "Hello World" to a global full-stack app. Python fans who want to learn about Cloudflare Workers should check it out (1/N) https://t.co/qTUtqNqG7y - **jarredsumner** (2025-12-12): https://x.com/jarredsumner/status/1999317065237512224 - Why is pasting into VSCode Terminal slow? Because it sleeps for 5ms every 50 characters. https://t.co/5ap5WajXN9 - **eastdakota** (2025-12-12): https://x.com/eastdakota/status/1999310673193697627 - @entemper Come back! Make it better!! - **thdxr** (2025-12-12): https://x.com/thdxr/status/1999296625236390378 - @goon_nguyen @opencode daily active users https://t.co/HsMD6itfHM - **adlonymous** (2025-12-12): https://x.com/adlonymous/status/1999289581209108520 - @irvinebroque Hey Brendan, I believe AI Agents should pay to consume certain npm packages. I'm building a pay-per-install package manager similar to Cloudflare's pay-per-crawl for AI agents (x402 ftw) - **rsms** (2025-12-12): https://x.com/rsms/status/1999275751053623533 - @raphaelschaad I like how there’s very little new to learn. The constraints are so tight that there’s little room for silly things like hover crap, home-made pop-up buttons, sidebars etc. - **zeeg** (2025-12-11): https://x.com/zeeg/status/1999256491367477669 - Who’s gonna tell him? https://t.co/OpaBytrrlU - **lessin** (2025-12-11): https://x.com/lessin/status/1999248291780002078 - The narrative-trust-capital flywheel... how it all works. https://t.co/jmCzQe9woW - **Yuchenj_UW** (2025-12-11): https://x.com/Yuchenj_UW/status/1999244943974629828 - I like this blog by @fidjissimo on OpenAI’s culture. "we’ve seen the biggest breakthroughs in AI begin as passion projects of a single researcher or two". GPT started as a passion project by Alec Radford, who doesn’t have a PhD. OpenAI feels like the old Bell Labs: bringing together geniuses and nerds (without requiring fancy schools and degrees), and giving them enough freedom to explore directions that matter but are not yet proven valuable. This may be the best model to run a research lab. - **fidjissimo** (2025-12-11): https://x.com/fidjissimo/status/1999235012160582125 - OpenAI turns 10 today. As a newcomer, I wrote about my observations of the very special culture that’s been built here: https://t.co/Rci9OCLp1a - **Cloudforce_One** (2025-12-11): https://x.com/Cloudforce_One/status/1999233469105238216 - UPDATE: Early activity indicates threat actors quickly integrated React2Shell into scanning routines, targeting critical infrastructure like nuclear fuel and uranium. React also disclosed two new vulnerabilities today—Cloudflare protects against all three. https://t.co/LSdi6XbuLe - **zeeg** (2025-12-11): https://x.com/zeeg/status/1999213281198616728 - need some of yall to stop doing startups and come help us build some things too many exciting things that we can bring to the masses, and not enough folks who want to be in the drivers seat - **zeeg** (2025-12-11): https://x.com/zeeg/status/1999196906883272970 - A while back we shipped a larger change in Sentry's MCP that I called Skills. This was an overhaul to our permission system that clustered tools around workflows less so than the upstream scopes they required. It was another move _away_ from it just being an API shim. - **thdxr** (2025-12-11): https://x.com/thdxr/status/1999185749007077393 - github is a very old app it has crazy wide scope, huge traffic, has to do some novel things its frontend architecture is all wrong lot of smart people trying to fix it even with a 5x smarter claude this isn't getting fixed much faster - **thorstenball** (2025-12-11): https://x.com/thorstenball/status/1999155279905402955 - The Amp CLI now has a map of your threads https://t.co/kEoJ47qOp8 https://t.co/weu3xgxBOR - **kate_rouch** (2025-12-11): https://x.com/kate_rouch/status/1999137349834289364 - Probably nothing https://t.co/dRCCYhRvHf - **KentonVarda** (2025-12-11): https://x.com/KentonVarda/status/1999110767736426528 - Lots of people saying that not all validation can be represented in TS types, e.g. "length < N" or "is a valid email". That's OK, I'm fine with writing code for those things. But 95% of the time I just want to verify that a string property is actually a string. - **raphaelschaad** (2025-12-11): https://x.com/raphaelschaad/status/1999104234432733460 - This is the most exciting trend in design. TUIs and GUIs are converging. While TUIs are getting more two-dimensional by the week, I suspect we'll see a trend of GUIs that are rendered (and work) more like modern-day TUIs. Can call that future style, "Tuimorphic." - **ankrgyl** (2025-12-11): https://x.com/ankrgyl/status/1999009971628028336 - people really underestimate how rapidly enterprises are adopting mcp - **tannerlinsley** (2025-12-11): https://x.com/tannerlinsley/status/1998962033224724706 - 👀 https://t.co/oDodWor7vj - **MiTypeScript** (2025-12-11): https://x.com/MiTypeScript/status/1998961519208595933 - ⚔️introducing TypeSlayer⚔️ A #typescript type performance benchmarking and analysis tool. A summation of everything learned from the benchmarking required to make the Doom project happen. It's got MCP support, Perfetto, Speedscope, Treemap, duplicate package detection, and more. https://t.co/qA1AyrqmaL - **WorkOS** (2025-12-11): https://x.com/WorkOS/status/1998957645689192598 - Rita Kozlov, VP AI and Agents at @Cloudflare , demos Code Mode, allowing agents to generate and execute JavaScript to improve tool calling performance, reduce token consumption and use native functions like Date to avoid hallucinations. An 81% reduction in token usage. #MCPNight - **KentonVarda** (2025-12-11): https://x.com/KentonVarda/status/1998937663995523205 - Folks stop explaining to me how type erasure works. I have designed and implemented erasure-based type systems before. There is no law that says you can't generate runtime code from a TypeScript type. Only self-imposed limitations. - **KentonVarda** (2025-12-11): https://x.com/KentonVarda/status/1998923888579981712 - No offense to the Zod devs but I don't understand why Zod has to exist. Why can't we just use the schema format we already have, that is, TypeScript? Why is there no way to construct a schema validator from a TypeScript type? This seems like such an obvious thing. - **gdb** (2025-12-10): https://x.com/gdb/status/1998897086079832513 - Agentic AI Foundation — advancing open-source agentic AI: https://t.co/5iGjhrMO0D - **dillon_mulroy** (2025-12-10): https://x.com/dillon_mulroy/status/1998884518959395179 - you can just ship prometheus exporters. now open source https://t.co/dB8LrGFg6b https://t.co/2UBZodiveQ https://t.co/gtIVdGY2nX - **brian_scanlan** (2025-12-10): https://x.com/brian_scanlan/status/1998849625537884532 - I did this instead: https://t.co/RTIwDsJ8NH https://t.co/vqWEeySSnc - **romainhuet** (2025-12-10): https://x.com/romainhuet/status/1998847480264557040 - Big step for open agentic AI. OpenAI is co-founding the Agentic AI Foundation under the Linux Foundation, alongside Anthropic and Block, and donating https://t.co/uop4XiFxan to this new umbrella. https://t.co/bV5XVEoJYh - **thdxr** (2025-12-10): https://x.com/thdxr/status/1998829022818578736 - it is crazy the amount of companies building on top of opencode - the built in webui is nice here and it's all with our low effort SDK i've been working here and there on making a higher effort SDK - hope to have it ready for everyone after the holidays https://t.co/yRLd0D4Ajh - **ghostwriternr** (2025-12-10): https://x.com/ghostwriternr/status/1998819156997579163 - In the next version of Cloudflare Sandbox SDK 📦 Running @opencode on remote sandboxes should become much easier. You can use the official OpenCode Typescript SDK to programmatically interact with the agent from your worker, or just expose the web UI and use it from there. https://t.co/IZYGOBbz1Y - **karpathy** (2025-12-10): https://x.com/karpathy/status/1998803709468487877 - Quick new post: Auto-grading decade-old Hacker News discussions with hindsight I took all the 930 frontpage Hacker News article+discussion of December 2015 and asked the GPT 5.1 Thinking API to do an in-hindsight analysis to identify the most/least prescient comments. This took ~3 hours to vibe code and ~1 hour and $60 to run. The idea was sparked by the HN article yesterday where Gemini 3 was asked to hallucinate the HN front page one decade forward. More generally: 1. in-hindsight analysis has always fascinated me as a way to train your forward prediction model so reading the results is really interesting and 2. it's worth contemplating what it looks like when LLM megaminds of the future can do this kind of work a lot cheaper, faster and better. Every single bit of information you contribute to the internet can (and probably will be) scrutinized in great detail if it is "free". Hence also my earlier tweet from a while back - "be good, future LLMs are watching". Congrats to the top 10 accounts pcwalton, tptacek, paulmd, cstross, greglindahl, moxie, hannob, 0xcde4c3db, Manishearth, and johncolanduoni - GPT 5.1 Thinking found your comments to be the most insightful and prescient of all comments of HN in December of 2015. Links: - A lot more detail in my blog post https://t.co/7LpJEVgbyk - GitHub repo of the project if you'd like to play https://t.co/WVQUbUzt2y - The actual results pages for your reading pleasure https://t.co/e2XIYElnc5 - **garrytan** (2025-12-10): https://x.com/garrytan/status/1998789947176992989 - People don’t want to buy software or platforms or even products They want to buy outcomes https://t.co/gnDAb8lYyl - **LevAkabas** (2025-12-10): https://x.com/LevAkabas/status/1998787916055851321 - For the second straight NBA season, the average 2 point shot is yielding more points than the average 3 point shot This is after decades of 3s being more valuable https://t.co/amzWsWXWgO - **bballinger** (2025-12-10): https://x.com/bballinger/status/1998786141395333445 - Proof small teams can win in AI: We trained a health foundation model on 3M person-days of data, comparable in scale to the larger labs. It detects high blood pressure with 87% accuracy, alongside atrial flutter (70%), ME/CFS (81%), and sick sinus syndrome (87%). The model was inspired by Yann LeCun's JEPA architecture, but adapted to multivariate irregular timeseries. We called it JETS: Joint Embedding for Time Series. The input is 63 channels of sensor data: heart rate, oxygen saturation, sleep stages, VO2Max, etc. The model was tested on both diagnosis and biomarker prediction tasks. The “AI revolution” won’t be limited to chatbots, images, or code assistants. We think physiological ground truth is the next frontier for health superintelligence. - **dok2001** (2025-12-09): https://x.com/dok2001/status/1998461003441266750 - We're excited to join the AAIF to continue @Cloudflare's work to support for open protocols for AI agents. The future of MCP is bright, and still only scratching the surface of what we think will be possible https://t.co/4v8azFVRzP - **mikeyk** (2025-12-09): https://x.com/mikeyk/status/1998456026136457532 - We're donating MCP to the @linuxfoundation and launching the Agentic AI Foundation with @OpenAI, @blocks, @AWS, @Bloomberg, @Cloudflare, @Google, and @Microsoft. MCP went from internal project to industry standard in a year. Now it gets the long-term stewardship it deserves. https://t.co/v3F5en5QkT - **Cloudflare** (2025-12-09): https://x.com/Cloudflare/status/1998450638590804043 - Cloudflare is joining the Agentic AI Foundation (AAIF) — the new home of the Model Context Protocol (MCP), https://t.co/h84Dk2PJOn and Goose. We're excited to continue our work to support open and interoperable standards for AI agents, to help build a better Internet for the AI era. - **eastdakota** (2025-12-09): https://x.com/eastdakota/status/1998444555931353124 - Excited to help define the future of the agentic Internet. https://t.co/0SEXwLw9O5 - **ritakozlov** (2025-12-09): https://x.com/ritakozlov/status/1998415599035470166 - you know you did something right when everyone starts copying you cloudflare has been offering a multi-tenant platform for platforms for years now, powering @shopify @hubspot @webflow and many others! allowing you to easily manage: 🔸 vanity custom-hostnames & tls certificate management 🔸 scalable multi-tenant compute with workers for platforms: including tagging, batch actions, per-script or tag limits, ability to run pre-worker or post-worker to customize outbound connections 🔸 easy per-tenant, horizontally scalable databases with d1 and durable objects 🔸all scaling down to zero, enabling you to pay for usage https://t.co/wjyfzxdfcd - **ankrgyl** (2025-12-09): https://x.com/ankrgyl/status/1998220957010571299 - the magic of simplicity is that you don’t ever think about all the work that went into solving the stuff you don’t experience - **dbreunig** (2025-12-08): https://x.com/dbreunig/status/1998120521247203356 - If this is true, it's the best argument that distribution and app design is more important than the model. Remember: MS has unfettered access to OpenAI technology. https://t.co/gEhpQDVysU - **MikeNomitch** (2025-12-08): https://x.com/MikeNomitch/status/1998107477540544861 - Python on Cloudflare Workers has gotten a lot better recently. Python is great... but loading packages into the top-level scope could be slooowwww. So now, Python Workers use memory snapshots to boot faster than Lambda and Cloud Run when using packages. More in thread... https://t.co/iOF4z4M7s4 - **iamdavidhill** (2025-12-08): https://x.com/iamdavidhill/status/1998006755133907403 - exploring subagent navigation ⦿ open sidebar with shortcut ⦿ search subagents ⦿ navigate between subagents ⦿ enter into a subagent https://t.co/jhHHLSR8XY - **jarredsumner** (2025-12-08): https://x.com/jarredsumner/status/1997914622477963537 - Bun's built-in PostgreSQL client is fast https://t.co/wRjrvx3C6p https://t.co/wib85IBefH - **dillon_mulroy** (2025-12-08): https://x.com/dillon_mulroy/status/1997893588555096178 - durable objects are for more than just websockets and multiplayer - they're an insanely good primitive for sharding, partitioning, and scaling compute https://t.co/51bxvOU5EO - **southpolesteve** (2025-12-08): https://x.com/southpolesteve/status/1997831048173129991 - What's the politest and kindest way to nudge an open source PR for a maintainer review. Really do think this will benefit a lot of zod users on workers. https://t.co/cTHusYVVpB - **thdxr** (2025-12-07): https://x.com/thdxr/status/1997815153384100164 - i still don't get hackathons - **nurodev** (2025-12-06): https://x.com/nurodev/status/1997403435998622049 - In case you needed a better way to manage scheduled tasks in Cloudflare Workers ⏲️ https://t.co/bSLTdTqhqO https://t.co/CvlnuB7Iyq - **elithrar** (2025-12-06): https://x.com/elithrar/status/1997392335152267491 - almost done: https://t.co/MML56OQTTb https://t.co/myqFoKQ4N9 - **infinterenders** (2025-12-06): https://x.com/infinterenders/status/1997359890021634509 - react fiber is honestly one of those things that people think they get until they actually look under the hood and then suddenly the whole thing feels like some alien technology living inside javascript. everyone parrots those same shallow lines like “fiber is the new diffing engine” or “fiber is concurrent rendering” and it’s just… no man, it’s so much deeper and way more deranged than that. the old react, the stack one, was like this innocent recursive tree walker that just kept diving into components like some happy kid climbing down a ladder, except that ladder was welded directly to the javascript engine’s callstack, so once you’re halfway down you can’t stop, you can’t turn around, you can’t take a break, you can’t do shit except finish it. so if you had big trees, like some giant UI with thousands of nodes, react would just lock the entire main thread and choke the browser until it was done. you typed into an input? too bad, the browser can’t react because react is too busy react-ing. it was legit painful. and the browser’s like “pls, bro, i need to paint” and react’s like “nah bro, i’m in the middle of rendering this dumb component”. so the react team basically looked at that mess around 2016 and realized the entire approach was screwed. like no amount of small optimizations could fix the fact that recursion is not interruptible in JS. the JS engine owns the stack, not you. and if you can’t pause, you can’t avoid jank, it’s just physics. so instead of fighting it, react just said screw the whole model and built their own stack. not metaphorically, literally. that’s what fiber is. it’s not some “algo”, it’s an entire fake callstack implemented as a linked structure of nodes. each fiber is like a tiny memory cell react controls completely, storing the component type, props, state, the children pointer, sibling pointer, parent pointer, priority info, effects flags, and all this other internal stuff. it’s like react turned every component into a little OS task. and the crazy part is that because it’s all controlled manually, react finally got freedom from the JS callstack. instead of recursive rendering, it’s this iterative stepping through fibers one by one. and that suddenly means react can pause. like just literally stop mid render, drop everything, let the browser repaint, handle input, chill, take a breath, then pick up exactly where it left off inside that huge UI tree. no more “oh no i’m stuck deep in recursion”. you can’t get stuck because react isn’t using the builtin stack anymore. it’s rolling its own. this is where the whole scheduling thing comes from. people think scheduling is some abstract idea but it’s literally react walking the fiber graph and deciding which updates matter more. typing is urgent. animations are medium. background data fetching is low. old react treated everything the same and that's why old apps felt like garbage when doing heavy work. fiber changed that whole world. suddenly react could say “yo, someone typed into an input, drop the low-priority render, resume later.” and the app stays smooth even if you have 50k components. and because fiber creates this double-tree thing, the current tree and the work-in-progress tree, react can run through the whole render phase in like a simulation mode. this is the wild part most ppl dont realize: when react is “rendering”, it’s not touching the dom at all. it's building this offscreen imaginary version of what the UI should look like. and only when the whole thing is done, consistent, no half-finished nodes, react switches to the commit phase and actually mutates the real DOM. like how game engines draw everything in a hidden buffer then flip it. that’s why the UI never shows broken half-updated states. fiber gives react the room to compute the perfect frame before actually showing it. and dude, this is exactly why suspense works. why transitions feel smooth. why server components can be streamed in and stitched into the client ig - **m2jr** (2025-12-06): https://x.com/m2jr/status/1997359479462203666 - This is one of the better takes on the AI "bubble" I've seen recently. https://t.co/sMJsm1vnT2 - **alexisxrivas** (2025-12-06): https://x.com/alexisxrivas/status/1997134211741110716 - Frank Gehry passed away today. Most people know him for the wild curves, crumpled paper inspiration, and metallic skins of his buildings, but there’s a more interesting story underneath. What most people don’t know is that his office was pushing form and fabrication so far that the existing tools of architecture simply broke. Conventional software couldn’t model it. Fabricators couldn’t price it or build it. Instead of pulling back, they built new tools. Gehry Technologies became the quiet enabler of that work. They adapted aerospace software (CATIA), built their own layers on top, and used it to coordinate geometry, structure, and fabrication with a level of precision the field had never seen. Behind the spectacle was process discipline. It proved that if you retool the process, you can make things that were previously impossible. Sometimes your only option is to invent the tools before you can build the thing itself. Whether or not you like every building he designed, Gehry forced the industry to evolve. And that’s a rare legacy. - **justinbieber** (2025-12-06): https://x.com/justinbieber/status/1997125645437194682 - If I hit this dictation button after sending a text and it beeps and stops my music one more time, I’m gonna find everyone at apple and put them in a rear naked choke hold Even if I turn off dictation I somehow hit the voice note thing The send button should not have multiple functions in the same spot - **hnshah** (2025-12-05): https://x.com/hnshah/status/1997062333500477483 - The real AI conflict shows up in Cloudflare’s logs. Four hundred sixteen billion crawler hits were stopped in a few months, and that single number tells you how quickly the web is starting to push back. Publishers spent years assuming crawlers were background noise. Now those requests carry real economic value because high-quality human content shapes how fast frontier models improve. The moment that value became obvious, the quiet pressure around access began to rise. Google fused its search and AI crawlers, which puts every creator in a complicated position. Stay visible in search and your work flows into training. Block the crawler and your visibility starts to fade. Cloudflare’s data shows how much tension that decision creates for the entire web. Look at the coverage numbers in the article. Google sees several times more of the internet than any other AI player. That reach creates a structural advantage that compounds as models absorb more of what only Google can see. Cloudflare is the rare company with enough placement on the internet to interrupt that advantage and give publishers a real choice. This story matters because it signals a new era. The open web is learning how to bargain. Infra companies are learning how to enforce those bargains. AI companies are learning that access is no longer guaranteed. The WIRED piece captures this moment before the new rules settle. It is worth your time because it shows where the leverage is shifting and why the next phase of AI development depends on negotiations that most people never see. https://t.co/yTMKcrwxJ7 - **yagiznizipli** (2025-12-05): https://x.com/yagiznizipli/status/1997015060343149011 - Some good news: I’m proud to say that I’m officially a @v8js committer! https://t.co/gxikaZoZ3R - **dok2001** (2025-12-05): https://x.com/dok2001/status/1996975716362305609 - You can read the details of what went wrong this morning in the post below. We let the Internet down again today; We’re locking down all changes in order to ensure we have better mitigation and rollback systems before we begin any again. https://t.co/oo1xJ2NsOK - **Cloudflare** (2025-12-05): https://x.com/Cloudflare/status/1996965903238299689 - Cloudflare experienced a significant traffic outage on December 5, 2025, starting approximately at 8:47 UTC, for approximately 25 minutes. We are sorry for the impact that it caused to our customers and the Internet. Here's what happened: https://t.co/FTJwncz36E - **samlambert** (2025-12-05): https://x.com/samlambert/status/1996759785468055677 - One day I will tell the story of Talib Kweli watching the GitHub infrastructure team remediate an incident. - **charlieholtz** (2025-12-05): https://x.com/charlieholtz/status/1996745328348070308 - How we built checkpointing And why we picked GPT-5 to do the hardest bits We recently added checkpointing to @conductor_build : in one click, you can reset your files, git, and chat to a prior state. Other dev tools have their own checkpointing features, but they're leaky. Suppose you ask Claude Code to write you a new feature. It installs a package, generates a database migration, writes some code, and stages its changes. Then you want to reset to before all that happened. You’ll see something like this: > git status Changes to be committed: modified: main.ts modified: migrations.ts new file: package-lock.json Changes not staged for commit: modified: main.ts The lockfile hasn’t been reset, migration hasn’t been reset and changes in main.ts have been reverted, but the old changes are still staged. This is because these checkpointing systems work by capturing the contents of a file right before the AI uses its file editing tool. That means it can’t capture changes that the agent makes through other means, like through the linter, package manager, code generator, etc. This might seem like a small problem, but we think it's rather big. It's like leaving a small gap in the fence at Jurassic Park: Conductor's job is to give you a contained environment for AI agents to run wild. A button that resets 90% of the AI's work is very different from a button that resets all of it! ## Exploring One simple thing we could do is create a commit for each checkpoint. But that would mess with your git history. We wanted it to feel like time travel, rewinding your chat history, files and git state as if by magic. Here are a few other ideas we had that didn't work out: - Commit to a private ref: Unfortunately, this misses uncommitted and untracked changes. We want to restore your untracked files, staged and unstaged changes, and commit history exactly as they were. - Store checkpoints using git stash: git stash create lets us stash without modifying files on disk, but it doesn’t let us include untracked files in the stash git stash -u && git stash apply would let us include untracked files in the stash, but it would modify files on the user’s local state temporarily each turn - Save the entire state ourselves in our sqlite db: This is a big undertaking. We’d have to rebuild functionality like git diff on our own. ## GPT-5 thinks At this point, we understood the system requirements well enough to write up an exact spec: https://t.co/XSDeWpN1rl We threw three things into GPT-5: - The above spec - A description of what we tried that didn't work (as above) - A short instruction: "Assess our options here. Tell me which direction you think we should go. Don't go deep into implementation details yet, but for any ideas you like, for each requirement, tell me in 1-2 sentences how that idea will meet that requirement. be careful not to miss any requirements" GPT-5 felt it could satisfy all the requirements using hidden git refs. Nice! But we didn't want to count any chickens until we had something to test. ## GPT-5 builds For building isolated subsystems, we've found that GPT-5 on its own often does a better job than coding agents like Claude Code or Codex. There's something about the coding agent environment that's distracting to the models when they're trying to build out something new. (I also get the sense GPT is a bit better at this kind of detail-oriented work than Claude is. For the same task today, I'd probably use GPT-5.1-Codex-Max.) I made two more requests to GPT-5: - can you sketch, for each function in the API, what the implementation would look like? ideally it'd be so that I can play around with it in my CLI by running the git commands directly, just to verify that it will work as intended - can you build a simple CLI tool called checkpointer with `checkpointer save`, `checkpointer restore`, and `checkpointer diff` options? And that was it! We tested checkpointer manually, then generated a test suite for it. Everything looked good. I don't think there were any major changes after that initial version, just a few tweaks to the API. GPT-5 virtually one-shotted the implementation. Here’s the final checkpointer script we're using today: https://t.co/7BJ3DkYENK ## How it works Our solution - Hooks into the agent’s lifecycle to run at the start and end of each turn - Captures three pieces of state: the current commit, the index (staged changes), and the worktree (all files including untracked). - Converts the index and worktree to tree objects using git write-tree—for the worktree, we write to a temporary index via GIT_INDEX_FILE first. - Bundles all three SHA-1s into a commit message and stores the commit as a private ref at .git/refs/conductor-checkpoints/. One downside of this approach is that if you have two agents running at the same time in the same workspace, you can’t unwind their changes separately; they’ll be all mixed together. That hasn’t been much of a problem for us, because we built Conductor to make it easy to create isolated workspaces where agents can run independently—and because we tend to use subagents to coordinate multiple agents working on the same task. Here's the checkpointer in action: - **mattzcarey** (2025-12-04): https://x.com/mattzcarey/status/1996719737749786846 - awesome!!! MCP Apps are gonna be huge. Someday we might even just call them Apps https://t.co/Mq0Kp7JBiO - **zachklein** (2025-12-04): https://x.com/zachklein/status/1996715864914284757 - Yes to more housing in the marina. Build everything Align proposes. https://t.co/TdNnjvFN1z https://t.co/dJslyiNwG2 - **dok2001** (2025-12-04): https://x.com/dok2001/status/1996653285160423889 - Please remember that a WAF rule is just a bandaid. Upgrade React today. There will be variations that of exploit that will need new protections. https://t.co/H7c7Aenc5p - **davidlee** (2025-12-04): https://x.com/davidlee/status/1996637188294156296 - @mhdempsey @jonwu_ wdyt of memory, battery requirements, data management? do you think these are the limiting factors to get to edge AI or model size? - **mattsimpsn** (2025-12-04): https://x.com/mattsimpsn/status/1996635861128200545 - super excited to share i’m joining @cloudflare! inspired by the incredible work, pace, and mission to build a better internet. grateful for everyone who made this possible and i can’t wait to get started 🧡 https://t.co/DP1zyW0Xu4 - **ritakozlov** (2025-12-04): https://x.com/ritakozlov/status/1996625162926735870 - i know i’m late to the party but opencode is extremely dope - **mhdempsey** (2025-12-04): https://x.com/mhdempsey/status/1996620143787376813 - On an enterprise level a large number of companies are not using the most frontier model today and the products work decently. Over time, I expect this to be even more true and is even where OAI is going with their routing of GPT-5. So right now based on improvement curve, yes, we only like frontier models, but as get better at scaffolding, tool use, etc. for a variety of applications I think it's possible we won't care about frontier model when flowed through a product UI/UX. - **ritakozlov** (2025-12-04): https://x.com/ritakozlov/status/1996603274724556860 - come watch me give a live demo of MCP + code mode at @WorkOS MCP night next week xoxo https://t.co/tDy8pmZBZP - **mhdempsey** (2025-12-04): https://x.com/mhdempsey/status/1996601661020643576 - the image on the left might be the most underappreciated chart in technology right now. https://t.co/RxtHlgaNIE https://t.co/nM8BLtZZhj - **grinich** (2025-12-04): https://x.com/grinich/status/1996600128275435656 - Have you seen "Code Mode?" Built by @cloudflare it allows you to convert MCP tools into TypeScript functions for more durable execution🤯 I'm thrilled to share that @ritakozlov will be speaking next week at MCP Night in SF. 🔥 RSVP now to save your spot! https://t.co/FkkWmJl4xk - **nxthompson** (2025-12-04): https://x.com/nxthompson/status/1996409716377854060 - Super interesting. What Anthropic found when it studied how its own employees are using AI. https://t.co/KF06UpH1Re https://t.co/rIG280tvkT - **dok2001** (2025-12-03): https://x.com/dok2001/status/1996264935907815829 - Shipping great technology is a team sport. As Cloudflare’s CTO I am surrounded by some of the most brilliant people I know. And I am looking for a unique person to join that group as Chief of Staff. The ideal candidate is a great communicator who can write code one minute and connect with customers the next. If you are interested, or know someone who is, please take a look at the job post! https://t.co/6tFFyFTzKZ - **dok2001** (2025-12-03): https://x.com/dok2001/status/1996247324901675296 - . @Cloudflare has deployed a new WAF rule to protect customers from a Remote Code Execution vuln that impacts React Server Components (CVE-2025-55182) used in frameworks like @nextjs . No action needed; the rule is enabled by default. You can learn more in our blog post https://t.co/XRgQOAyun1 - **thdxr** (2025-12-03): https://x.com/thdxr/status/1996058236605399211 - the crazy part is this respects opencode's client/server model so even if opencode server is running on a different machine the shell session will run there and stream the results to the client incredible for sandboxes https://t.co/B86g4LFr5d - **elithrar** (2025-12-03): https://x.com/elithrar/status/1996032489954181301 - installed @opencode on my personal repos so I can PR fixes + deploy them to Workers from the sofa 🙃 https://t.co/WdE3BCuer9 - **zeeg** (2025-12-02): https://x.com/zeeg/status/1995960329600471310 - re:invent is the conf where SF flies to Vegas to schedule a bunch of networking events with people (also from SF) that dont actually want to spend time with with them, at places they dont actually want to visit - **GrantSlatton** (2025-12-02): https://x.com/GrantSlatton/status/1995897289152627076 - i've spent the last decade mainly reading good technical writing it's usually written in a style where each paragraph is short and conveys a single idea, usually in 2-4 sentences this has rendered me incapable of parsing old literature with its half-page-long paragraphs - **MarcSelwan** (2025-12-02): https://x.com/MarcSelwan/status/1995657772739010847 - If I was hiring product manager roles for developer centric products today, I think I’d put more of an emphasis on knowing how to understand code now more than ever. No, I’m not talking about leet code, I’m talking about reading code and understanding what it does. Why? Because we have more power than ever to quickly prototype, test, fix bugs, and add simple features in the advent of coding assistants. This is effectively a measure of how much friction you can reduce for the engineering teams you work with. - **sherwinwu** (2025-12-01): https://x.com/sherwinwu/status/1995635733844128026 - Had a blast talking with @martin_casado about the OpenAI API — model customization, fine tuning, pricing, open weights, and the agent builder! https://t.co/3gh626Fry4 - **mwylde** (2025-12-01): https://x.com/mwylde/status/1995559903738691590 - But that's not all—we've also added a complete Apache Iceberg sink connector, making Arroyo the easiest way to process and ingest streaming data to Iceberg tables. There's way too much in this release to tweet it all, so check out the release notes: https://t.co/A1jZ06ya6P - **dillon_mulroy** (2025-12-01): https://x.com/dillon_mulroy/status/1995548739293831438 - ngl having a ton of fun building w/ hono and workers - **Cloudflare** (2025-12-01): https://x.com/Cloudflare/status/1995512869966811293 - Today, we’re excited to announce that Replicate is officially part of Cloudflare. We wanted to share a bit about our journey and why we made this decision. https://t.co/inEXJvwUsw - **firstadopter** (2025-11-30): https://x.com/firstadopter/status/1995191128321048784 - Here's what's going to happen from here: -Nvidia will report blowout numbers the next several quarters on the NVL72 product supercycle (a step function up in capability with 72 GPUs in one AI server versus 8 GPUs). It will become one of the largest cycles in technology history, akin to the iPhone versus Blackberry. The clear signs are there in the latest quarter with Nvidia posting its first accelerating revenue growth in two years and triple-digit networking segment growth. -Google TPUs will be less than 10% of the market for the next few years as the major hyperscaler buyers don't want to support a cloud rival (outside of Meta), while Nvidia's software CUDA/developer ecosystem, TSMC allocation and performance advantages remain strong. -Most search queries will transition to AI chatbots over the next few years. -ChatGPT will release a much better model trained on NVL72 Blackwell clusters at Microsoft. Sentiment will shift back to ChatGPT. -ChatGPT will add digital advertising in its consumer product. The first iteration will not be great. The second version will improve. The third version will work great. -A significant portion of the digital ad market will move to AI chatbots and AI consumer hardware, away from search ads. -Google will go from 95% search monopolist to a number 2 or number 3 player in the AI chatbot market, which will dramatically lower its margins over the cycle. Serving the search index was a gold mine. That era will end. Timing is difficult. It may take a while, but it will happen. -All the talk about AGI and superintelligence is a distraction and a side show. AI adoption and AI progress will accelerate through 2026 as Gemini and Claude Opus proved scaling laws are intact. Enterprises will unleash massive productivity gains using current technology. Cursor will be the precursor (get it?) of the future. It eliminates tedious work with autocomplete, bug fixing, leading to rapid iteration of new ideas for coding. It enables 40% more productivity. There will be a Cursor for every vertical. Knowledge workers will become vastly more productive as AI models build upon intuitive understanding of what helps them with proprietary custom data and models. -But don't they lose money now? Compute performance continues to improve and costs will come down. This is inevitable based on history. Today's loss-making features will become enormously profitable in due time. Curious to hear your thoughts. - **jam_croissant** (2025-11-30): https://x.com/jam_croissant/status/1995169213225328767 - As @profplum99’s recent paper (& an ♾️# of academic studies show) #’s like CPI doesn’t account for the difference in inflation across different socioeconomic classes… Lower-income households face significantly higher inflation than rich ones, on the order of ~0.7% per year. Taking 1 inflation number, applying it across all socioeconomic classes equally & then not accounting for the compounding of this effect over 58 years, is propaganda at its finest. 👇 👇 👇 👇 👇 👇 👇 👇 👇 👇 - **_chenglou** (2025-11-30): https://x.com/_chenglou/status/1994952618737868867 - Explanations of Attention in ml are so complicated. I think of it as roughly a hash table lookup, except instead of using an exact key, you use a fuzzy one (query) to retrieve all related keys and average their values (A weighted average, where each weight is the query and that key’s affinity) - **thdxr** (2025-11-29): https://x.com/thdxr/status/1994784992598032522 - i wish people understood both of these things 1. yes software quality is generally pretty bad and we need to figure out how to try harder 2. there is so much software that needs to be built - making it more accessible with higher level languages, frameworks, AI, services, is a good thing - **matthewcp** (2025-11-28): https://x.com/matthewcp/status/1994455536910168437 - "no AI PRs" is really silly. What's next, you're going to tell people what editor they can use? The submitter is responsible for the code they are asking for review. What tool made the edits is irrelevant. - **BraydenWilmoth** (2025-11-28): https://x.com/BraydenWilmoth/status/1994429576081252487 - I've been writing about this a lot internally lately. How teams work is changing. Reducing the gaps between designers → engineers → users with AI will be a competitive advantage. Teams that historically did not collaborate directly, can now work together seamlessly. https://t.co/LK9plKmFPT - **momito** (2025-11-27): https://x.com/momito/status/1994109226915078583 - i've been thinking about how to make the cloudflare primitives visuals more useful so i built a CLI that turns each visual into a simple starter template you can deploy inspired by a convo w/@irvinebroque earlier this week repo: https://t.co/FBKLKJ8Lj3 - **BraydenWilmoth** (2025-11-27): https://x.com/BraydenWilmoth/status/1994017075350716486 - @syeefk @Cloudflare @burcs I removed roughly 16 of these this week that should go live next Tuesday. 😊 - **grinich** (2025-11-27): https://x.com/grinich/status/1993884237318246509 - the single best thing about @johncoogan and @jordihays is that they are optimists. they want the world to be better and they want you to win. the traditional media seems to just chase clicks through negativity and tearing you down. @tbpn in contrast is about celebrating success, being excited about the future, and having fun along the way. it’s a simple recipe and a breath of fresh air. it’s also the kind of energy that made me want to start a startup almost 15 years ago. the world needs more optimists. i can’t wait to go back on the show myself. - **samlambert** (2025-11-27): https://x.com/samlambert/status/1993838723793080800 - Discussing a product launch in Slack and @ mentioning @cursor to read the conversation and have an agent write the blog post and docs is a wild workflow. - **skeptrune** (2025-11-26): https://x.com/skeptrune/status/1993752697603018981 - han and i had a long back and forth about saying "faster page loads" since we improved TTFB rather than content download time😜. you can tell who won lol. we shipped a new solution on top of Cloudflare for this release which reduces TTFB (time to first byte) in the P75+ case by getting our cache hit rate to 98% from its previous 76% across all sites we host our pages are still heavy and we have much work left to do there, but all of the loads at least feel fast now since the TTFB P99 is now <100ms. also, Vercel shipped a shield solution which replicates what we scaffolded together ourselves on Cloudflare and we are going to be both trying that and yapping about out it in our eng blog soon! - **MikeNomitch** (2025-11-26): https://x.com/MikeNomitch/status/1993533303606198736 - Cloudflare Containers, now with mountable R2-backed volumes - thanks to newly added FUSE support *Easier interface for setting these up coming soon, but not bad to set up today following Matt's linked example https://t.co/iL2rXcv4FA - **momito** (2025-11-25): https://x.com/momito/status/1993363147693212104 - next up: hyperdrive https://t.co/VvxaeH8t3n https://t.co/cETEFFD7rg - **elithrar** (2025-11-25): https://x.com/elithrar/status/1993326358173192369 - @nandafyi needs to communicate the current account more clearly (since that is critical context when acting on resources). switcher itself is great. needs a shortcut (⌘A) to open as well. - **swyx** (2025-11-25): https://x.com/swyx/status/1993263990391378152 - words cannot express how based i find the @AmpCode experiment with ad driven inference. rumored to be $5-$10m ARR already lmao. they got here before openai, open router, ollama or any of the other supposedly innovative coding companies. its the second oldest internet business model. it’s not that i think this specific thing will scale to the moon. it’s more that the nimbleness of @beyang and @sqs as a >10 year old team going founder mode is extremely positive for wherever they end up landing as a business. - **jam_croissant** (2025-11-25): https://x.com/jam_croissant/status/1993196531848298776 - Nixon 1.0 vs. Nixon 2.0 👇 👇 👇 Trump 2.0 …. Coming to a presidency near you. https://t.co/YYSAQMDDTJ - **bradgarropy** (2025-11-25): https://x.com/bradgarropy/status/1993169526960332856 - @jamesperkins @Cloudflare Alright, it's a bug introduced after we upgraded to React 18. Safari's refs to the side modal aren't updating before the useEffect focus trap is running. Working on a solution. https://t.co/ZWpkYbhTeQ - **thdxr** (2025-11-25): https://x.com/thdxr/status/1993139896228495516 - bunch of people are asking me for my reaction about anthropic saying "software is done in 6 months" or whatever here's my response https://t.co/DxY2Um13zZ - **yagiznizipli** (2025-11-25): https://x.com/yagiznizipli/status/1993121152768921901 - Excited to share that @erikcorry, @lemire, and I improved the performance of TextEncoder encodeInto() in Node.js by 12-1984%. This work was made possible by @Cloudflare’s dedication to open source https://t.co/GhZILZ5040 - **nandafyi** (2025-11-24): https://x.com/nandafyi/status/1993043615913066859 - account switcher https://t.co/rTqf4SZsUB - **abhiarya** (2025-11-24): https://x.com/abhiarya/status/1993033150323392720 - https://t.co/I99HkPgPku - **_irtefa** (2025-11-24): https://x.com/_irtefa/status/1992974833441739071 - I'm heading back to Cloudflare. It was an easy decision. It’s no secret that @eastdakota and @zatlyn have built a unique company, and I couldn’t pass up the chance to work with @dok2001 and @LakeAustinBlvd again. My first stretch at Cloudflare left a mark. A few memories I still think about: - The Austin office where the whole team fit inside a tiny WeWork - Helping ship Cloudflare Workers to the world for the first time - Watching Cloudflare ring the bell and go public - Helping ship Cloudflare’s first mobile app It’s a special time at Cloudflare. I’m thrilled to be back and ready to create some new memories. - **swyx** (2025-11-23): https://x.com/swyx/status/1992681925308260688 - very honored to briefly bring @thdxr out of paternity leave for an incredible online track keynote from one of the most transparent, brutally honest builders in our space! taken in one shot, no less. the casual flex... https://t.co/vCkLbeoxx5 https://t.co/J6Qq8kpah9 - **btaylor** (2025-11-23): https://x.com/btaylor/status/1992645831464280560 - I am an Inbox Zero guy. My workflow relies on each email having three states (read, unread, archived) because I often read / triage / archive before answering. Has anyone figured out a similar workflow with Apple Messages? Marking unread doesn’t cut it because it breaks triage - **JesseJenkins** (2025-11-23): https://x.com/JesseJenkins/status/1992456609617076505 - Rooftop solar in America is unacceptably expensive. https://t.co/jX652sSI1E - **elocinationn** (2025-11-22): https://x.com/elocinationn/status/1992280688968765886 - There’s a belief in SF that with enough money you can solve anything. For a lot of technical companies, VCs are therefore looking for a charismatic founder who they believe will be able to get that money. In the first few rounds, they care less about the actual technical progress. VCs won’t directly tell you this but it is true, or at least revealed through their choices of who to fund. You’re seeing companies winning big rounds with no tech, followed by a scramble to acquire the people to build the tech for them. The problem with this is the word spreads quickly that you are a bit of a bullshit company amongst the people you want to hire. I’ve heard accumulating insider stuff lately that makes me believe this has gone too far. Money going to young founders or founders with no domain knowledge because they told a great story, but then technical friends go to interview there and are horrified. This is especially true on the life science side of things, where biologists are going into labs and seeing what looks like a 1st year try to do experiments and making obvious critical mistakes. Next the founder will argue with said interviewee, insisting they’re wrong only to then ask chatGPT in front of them and further humiliate themselves. Interviewee then refuses to join, and slowly finds others in their network who have done the same. Now it’s even harder for this founder to recruit anyone serious to solve their problem and they panic. Next they start to fabricate data and marketing materials. We all know where this goes. There seems to be a new bio wave coming through atm (imo GLP-1 success has renewed belief in power law outcomes). I think the challenge with life sciences in particular is the people who go into it are less financially motivated, and so to recruit them, you genuinely need to be doing excellent work and know your stuff. VCs with limited exposure to scientists may be making an error in thinking everyone has the same motivations as the average software engineer. Winning over talent for your life science company is different from software. Arguing with someone with deep technical expertise and being wrong is extremely bad for the company’s reputation and spreads fast. All this is to say is be careful. Many of these founders seems to sell a story that they’re going to do things faster and better than trad bio, but once you get in there they’re actually just doing the same thing in a much less sophisticated way. There is basically nothing new. It’s like you’re giving them money for a science project and to do a course on how to be a research assistant in a lab. For anything that’s science problem-like, make sure the founding team has someone who is respected in their field. Maybe get someone from your network with expertise to step into a lab with them and check. Granted being able to fundraise is important, this blanket rule isn’t translating as well to life sciences. They also need to be able to hire the best people. - **fidjissimo** (2025-11-22): https://x.com/fidjissimo/status/1992136485131506128 - We need the equivalent of the 20th-century clean water revolution, except for clean indoor air. People with post-infectious chronic illnesses have known for a while how insane it is that, as a society, we continue to find it perfectly acceptable to breathe air full of viruses that can completely destroy our lives years later and that our schools are virus factories. With more horrible conditions being attributed to viruses earlier in life, I hope that perceptions will finally change, and that contaminated air will soon feel as unacceptable as contaminated water. - **MikeNomitch** (2025-11-21): https://x.com/MikeNomitch/status/1991953734386295169 - We're not really worried about idle containers CPU-wise. Memory and disk are still priced at the provisioned amount (I can follow up on the main thread be explicit here) - so incentives are aligned where idle containers are okay and you aren't crowding out other workloads "for free" - **MikeNomitch** (2025-11-21): https://x.com/MikeNomitch/status/1991938274760462730 - Cloudflare Containers and Sandboxes prices just got a lot better. You’re now billed on actual CPU usage, not the CPU reserved for an instance. If you only use 20% of your vCPU capacity on average, you pay 5X less for CPU time. - **ritakozlov** (2025-11-21): https://x.com/ritakozlov/status/1991935594898116636 - 🚨 we’ve updated @cloudflare containers pricing (and thus sandboxes) to only charge for usage, not allocation this is going to make a huge difference especially for sandbox execution for agentic use cases https://t.co/m662q5xUDU https://t.co/oVGVZ2esgQ - **matt_slotnick** (2025-11-21): https://x.com/matt_slotnick/status/1991926714289840628 - that is to say that i wonder if the enterprise is significantly underserved by the current generation of labs, and thus underserved by the majority of capital being deployed towards AI - **BraydenWilmoth** (2025-11-21): https://x.com/BraydenWilmoth/status/1991902194640302430 - Introducing a brand new Workers create flow coming soon, but why not give you all a first peek at what we've been cooking up. Coming from chef @hkfoster 🧑‍🍳 https://t.co/MJc3pDkXYb - **KentonVarda** (2025-11-21): https://x.com/KentonVarda/status/1991889320308084760 - One of these days someone is going to create affordable hardware to run supersize AI models (e.g. Kimi K2) for *one user at a time*, and they are going to disrupt the hell out of NVidia. Like textbook Clay Christensen style. - **brian_lovin** (2025-11-20): https://x.com/brian_lovin/status/1991643111307767897 - Wake up babe, new S-tier personal site just dropped https://t.co/stGwRDRLM2 - **samlambert** (2025-11-20): https://x.com/samlambert/status/1991631975329984682 - .@intercom serve 10M QPS from IdentityCache https://t.co/l1NSeMwUjB - **mjackson** (2025-11-20): https://x.com/mjackson/status/1991612529760694523 - If you think this is cool, you're going to be blown away once you start digging into how much work we've done on the server side in Remix 3. We've agreed to not talk about it too much until we're ready to unveil the full picture, but everything is being done on GitHub. https://t.co/Jkl8Hll0ET - **pitdesi** (2025-11-20): https://x.com/pitdesi/status/1991547299856805961 - How can you not love this guy https://t.co/sAMTnPlwL9 - **Birdyword** (2025-11-20): https://x.com/Birdyword/status/1991523049737449684 - This week I wrote about the de-mortgaging of America. As it becomes increasingly difficult to buy a home, American mortgage debt is falling - to a 25 year low relative to GDP, and a 60+ (!!!) year low relative to the value of the housing itself https://t.co/ozD8MrEfMF - **zebassembly** (2025-11-20): https://x.com/zebassembly/status/1991520452209176926 - This is SO good, definitely linking to this when people ask what are Durable Objects. Makes me wonder if we should have animated diagrams in our docs https://t.co/v8JkzEpsDf - **boristane** (2025-11-20): https://x.com/boristane/status/1991510565433536884 - this is nice https://t.co/rvCgam6vOM - **ryanflorence** (2025-11-19): https://x.com/ryanflorence/status/1991282288844501249 - See that lazy-file package? You can "open" files without reading them, then just throw them at a response and keep your memory footprint itty-bitty There is so much good stuff behind all of this, and it's all built on standards, so this same server could run in a service worker - **ryanflorence** (2025-11-19): https://x.com/ryanflorence/status/1991281680976658797 - I'm in love. A JavaScript server that serves static files out of a public directory and routes everything else to an index.html SPA style https://t.co/1qb1PoViYK - **michellechen** (2025-11-19): https://x.com/michellechen/status/1991225051421561134 - ai gateway unified billing is rolling out to everyone who signed up for the closed beta just in time for people to try out Gemini 3 via AI Gateway! just a few clicks to load money into your AI Gateway wallet and start using top models from Google, OpenAI, Anthropic and more - **terronk** (2025-11-19): https://x.com/terronk/status/1991201513457193284 - @0thernet @zocomputer I love that your mom test is not “so easy my mom can use it” and actually “so powerful my bio researcher mom and her grad students can use it.” Need more mom tests like this. - **michaeljburry** (2025-11-19): https://x.com/michaeljburry/status/1991194519090368702 - Early is not wrong. https://t.co/9jsyJYOGZS - **tracyalloway** (2025-11-19): https://x.com/tracyalloway/status/1991129738509730082 - IMHO - arguing over whether Gen Z is better or worse off economically *right now* than previous generations is useless and misses the point since most of the angst is the belief that their economic *future* is worse (and no one can disprove that yet) - **zebassembly** (2025-11-19): https://x.com/zebassembly/status/1990970833494233323 - Logs Written in the Workers Logs part is the thing that should indicate that we only charge for logs persisted and not exported to an otel destination. cc @nevikashah should we adjust the wording in the docs? - **southpolesteve** (2025-11-19): https://x.com/southpolesteve/status/1990962207279009871 - Heard that zod can be slow on Cloudflare Workers. Let's fix that https://t.co/QMhgWzLaPg - **boristane** (2025-11-18): https://x.com/boristane/status/1990931658699403642 - amp's oracle and librarian are truly good ideas - **eastdakota** (2025-11-18): https://x.com/eastdakota/status/1990931568785854704 - This is a good point. https://t.co/PsixcbchZG - **zebassembly** (2025-11-18): https://x.com/zebassembly/status/1990928010229293482 - @SamNewby_ Use Workers Observability with an OTLP destination to export to a separate provider https://t.co/UQakr5GSdK - **zebassembly** (2025-11-18): https://x.com/zebassembly/status/1990925597015494931 - Workers Observability! It's a structured logging/tracing solution built into the platform that can be queried in the dashboard or be configured to export to a third party Observability platform via OpenTelemetry! We also have an investigate tab where you can visualize things based off your logs. - **eastdakota** (2025-11-18): https://x.com/eastdakota/status/1990925549892456558 - We let the Internet down today. Here’s our technical post mortem on what happened. On behalf of the entire @Cloudflare team, I’m sorry. https://t.co/uTi23Vcx0g - **zebassembly** (2025-11-18): https://x.com/zebassembly/status/1990876489101815943 - The "don't return two nils" convention in Go is stupid. I'm fallibly returning an optional, why the hell would I want to make the code more complex with a special marker error. Can't stand this language - **zeeg** (2025-11-18): https://x.com/zeeg/status/1990840161211724035 - This take is spot on. Hiring has not slowed, but what has is peoples desire to succeed. I see a lot of behaviors from my position at Sentry, and one common behavior is people phoning it in. I can't tell you why, but I can tell you its impossible to hide. https://t.co/Ac6sVrLILd - **dok2001** (2025-11-18): https://x.com/dok2001/status/1990827571089428580 - Sharing an update on the recovery of our services. We were able to resolve the impact to traffic flowing through our network at approximately 14:30 UTC, which was our first priority, but the incident required some additional work to fully restore our control plane (our dashboard and the APIs our customers use to configure Cloudflare). The control plane should now be fully available. We are monitoring those services and continuing to ensure that everything is fully operational. Again, we plan to share a complete walkthrough of what went wrong today in a couple of hours and how we plan to make sure this never happens again. - **dok2001** (2025-11-18): https://x.com/dok2001/status/1990791419653484646 - I won’t mince words: earlier today we failed our customers and the broader Internet when a problem in @Cloudflare network impacted large amounts of traffic that rely on us. The sites, businesses, and organizations that rely on Cloudflare depend on us being available and I apologize for the impact that we caused. Transparency about what happened matters, and we plan to share a breakdown with more details in a few hours. In short, a latent bug in a service underpinning our bot mitigation capability started to crash after a routine configuration change we made. That cascaded into a broad degradation to our network and other services. This was not an attack. That issue, impact it caused, and time to resolution is unacceptable. Work is already underway to make sure it does not happen again, but I know it caused real pain today. The trust our customers place in us is what we value the most and we are going to do what it takes to earn that back. - **TheRohanVarma** (2025-11-18): https://x.com/TheRohanVarma/status/1990677046864064911 - I become Cursor’s first full-time PM 4 months ago. What I am NOT doing: - Manage a product sprint - Grooming - PRD What I am doing: - Help build, price, market, and launch our first second product: Bugbot - Customer support for Bugbot for the first couple months - I’ve responded to literally thousands of emails - Enable and help our rapidly scaling GTM to sell Bugbot and Cloud agents to enterprise - Spend lots of time with our biggest customers who are least similar to our engineering team - Talk to hundreds of users for competitive analysis and UXR - Advocate for and help kick off some of the features that became Cursor 2.0 (agent layout + plan mode) - Architect and scope a new platform product we’ll be releasing soon - Help GTM close deals and expand our footprint in our enterprise contracts directly in the sales process - Travel across the world to visit our customers, help with enablement, and see where we can improve - Helping with misc random things: evaluating strategic acquisitions, “thought leadership” events, lots of recruiting, product/content marketing, devrel, and more Am I PM-ing correctly? 🤷🏾‍♂️ - **zeeg** (2025-11-18): https://x.com/zeeg/status/1990616714674082105 - Can we all agree that the current AI wave has made “devrel” / “influencers” (aka the uniformed) more annoying than ever? Go back to building, learn things, have real discourse, and ignore all the noise. - **ritakozlov** (2025-11-17): https://x.com/ritakozlov/status/1990532000227201469 - cloudflare nyc office golden hour is simply unmatched https://t.co/AEioPlVp0N - **brezina** (2025-11-17): https://x.com/brezina/status/1990500223169540512 - @harris Meh. Much better to impress or scare the CEO. Ideally both - **miacmal** (2025-11-17): https://x.com/miacmal/status/1990498841058254985 - I’ll start 🙋‍♀️ - Surfacing queue backlog count in Workers consumers - Notifications on queue lag - Logs directly in Queues dash https://t.co/O0gD3ZdScF - **miacmal** (2025-11-17): https://x.com/miacmal/status/1990492959528321475 - happy monday! ☀️ we’re working on improvements around @Cloudflare Queues observability. if you have any feature requests around this, let me know! could be anything from notifications to graphql to dashboard… whatever comes to mind 👀 - **BornsteinMatt** (2025-11-17): https://x.com/BornsteinMatt/status/1990484310873514235 - Congratulations to @bfirsh, @allnoteson and the @replicate team!! You built a foundational infra product in AI long before the tech was well understood. And you influenced a whole generation of builders / how we build with AI. While it's bittersweet to see the company get acquired, @Cloudflare is the one of the few public companies that embodies the same product ethos & love for developers as you all do - **chrismunns** (2025-11-17): https://x.com/chrismunns/status/1990472601471480230 - It feels like there's a fairly big line in the sand being drawn between "legacy" cloud, for old school enterprise apps/COTS, and new-modern cloud for AI apps, SaaS, the future. The industry will need both, but I absolutely know where i'd go to build new, and where i'd go to run old. And they are not the same place. And that gap is widening rapidly. - **wesbos** (2025-11-17): https://x.com/wesbos/status/1990450743535702232 - this is fantastic news. I love both these companies https://t.co/jyYk4XLq87 - **corywilkerson** (2025-11-17): https://x.com/corywilkerson/status/1990448380049605088 - The best dev tools on the planet are being built at @cloudflare. If you know, you know. Can't wait to be part of the team and bring @replicate into its next and best form yet. Big things coming. 🧡 https://t.co/IgEkDw334q - **elithrar** (2025-11-17): https://x.com/elithrar/status/1990442384421749093 - mount an R2 bucket inside a @Cloudflare Container: https://t.co/s6IEHJcui3 — and use it to bootstrap your sandboxes, long running processes and tools that don't speak the S3 API. real docs and a much more polished experience coming soon. https://t.co/MfTrs5KgYF https://t.co/bwT9oezFEo - **minglu** (2025-11-17): https://x.com/minglu/status/1990429771017470013 - The default way to build AI apps: https://t.co/xo3yQbKDxw - **GrantSlatton** (2025-11-17): https://x.com/GrantSlatton/status/1990429465101447337 - most of my personal-use AI tools run on Cloudflare workers and call out to Replicate APIs great to have this in one place, seems like an ideal infrastructure/vision match too https://t.co/oUGmVJAuY1 https://t.co/ujP5Ysy2Bl - **acoyfellow** (2025-11-17): https://x.com/acoyfellow/status/1990426252360941898 - Getting ready for a @inboxdog demo later today... Here it is one-shotting a full self-learning email assistant in a few words: https://t.co/jvBtERd64T - **Cloudflare** (2025-11-17): https://x.com/Cloudflare/status/1990425945195389194 - Big news! https://t.co/EMXILJEJU5 https://t.co/sQhngGx73j - **replicate** (2025-11-17): https://x.com/replicate/status/1990424622165328095 - Big news: Replicate is joining Cloudflare. Replicate's going to carry on as a distinct brand, and all that will happen is that it’s going to get way better: it’ll be faster, we’ll have more resources, and it’ll integrate with the rest of Cloudflare’s Developer Platform. https://t.co/BOACiNGORH - **mdhruvill** (2025-11-16): https://x.com/mdhruvill/status/1990152234513182812 - Introducing GitVex - a fully open-source serverless git hosting platform. No VMs, No Containers, Just Durable Objects and Convex. Code: https://t.co/bZBotx6kUC Built using: @Cloudflare @tan_stack @convex @coderabbitai @Sentry Demo in reply 👇 https://t.co/yrrIozP6dC - **danshipper** (2025-11-16): https://x.com/danshipper/status/1990141001025913187 - this is extremely true and under discussed https://t.co/sGWQKv7WeI - **karpathy** (2025-11-16): https://x.com/karpathy/status/1990116666194456651 - Sharing an interesting recent conversation on AI's impact on the economy. AI has been compared to various historical precedents: electricity, industrial revolution, etc., I think the strongest analogy is that of AI as a new computing paradigm (Software 2.0) because both are fundamentally about the automation of digital information processing. If you were to forecast the impact of computing on the job market in ~1980s, the most predictive feature of a task/job you'd look at is to what extent the algorithm of it is fixed, i.e. are you just mechanically transforming information according to rote, easy to specify rules (e.g. typing, bookkeeping, human calculators, etc.)? Back then, this was the class of programs that the computing capability of that era allowed us to write (by hand, manually). With AI now, we are able to write new programs that we could never hope to write by hand before. We do it by specifying objectives (e.g. classification accuracy, reward functions), and we search the program space via gradient descent to find neural networks that work well against that objective. This is my Software 2.0 blog post from a while ago. In this new programming paradigm then, the new most predictive feature to look at is verifiability. If a task/job is verifiable, then it is optimizable directly or via reinforcement learning, and a neural net can be trained to work extremely well. It's about to what extent an AI can "practice" something. The environment has to be resettable (you can start a new attempt), efficient (a lot attempts can be made), and rewardable (there is some automated process to reward any specific attempt that was made). The more a task/job is verifiable, the more amenable it is to automation in the new programming paradigm. If it is not verifiable, it has to fall out from neural net magic of generalization fingers crossed, or via weaker means like imitation. This is what's driving the "jagged" frontier of progress in LLMs. Tasks that are verifiable progress rapidly, including possibly beyond the ability of top experts (e.g. math, code, amount of time spent watching videos, anything that looks like puzzles with correct answers), while many others lag by comparison (creative, strategic, tasks that combine real-world knowledge, state, context and common sense). Software 1.0 easily automates what you can specify. Software 2.0 easily automates what you can verify. - **OutwestLock** (2025-11-15): https://x.com/OutwestLock/status/1989802616948470209 - That boy 30 crazy as hell https://t.co/Vwomyq204m https://t.co/RYY2aWNjiS - **dbroockman** (2025-11-15): https://x.com/dbroockman/status/1989762118946886064 - The last three dinner parties I’ve been at in SF, with completely non-political people, have all at some point descended into everyone exclaiming some version of “so thankful for @DanielLurie” - **ericclemmons** (2025-11-15): https://x.com/ericclemmons/status/1989490376198242503 - Ok, this is the coolest thing I've built YET with @Cloudflare 😍 No-code Playwright E2E Testing + CI ⏱️ 30s from Prompt to Automated Test ✅ 90s E2E Test CI It's basically an uptime checker for products 💪 https://t.co/cHRs7gUDlg - **samlambert** (2025-11-14): https://x.com/samlambert/status/1989186129451708639 - I don't get the Mongo hate. Mongo had replication before Postgres and their v1 of replication was better than Postgres' is today. - **zebassembly** (2025-11-14): https://x.com/zebassembly/status/1989160813500985783 - it's really hard to be on this website as just a programmer and see all the vile shit other people say. I don't like that my timeline is programming - programming - post with horrifically racist or transphobic relies - programming. - **karpathy** (2025-11-13): https://x.com/karpathy/status/1989078861800411219 - I am unreasonably excited about self-driving. It will be the first technology in many decades to visibly terraform outdoor physical spaces and way of life. Less parked cars. Less parking lots. Much greater safety for people in and out of cars. Less noise pollution. More space reclaimed for humans. Human brain cycles and attention capital freed up from “lane following” to other pursuits. Cheaper, faster, programmable delivery of physical items and goods. It won’t happen overnight but there will be the era before and the era after. - **samgoodwin89** (2025-11-13): https://x.com/samgoodwin89/status/1989053062946910236 - Effect is making me fearless. Alchemy’s new providers are more reliable and take less effort to build. The compiler catches all the edge cases. They write themselves. - **dillon_mulroy** (2025-11-13): https://x.com/dillon_mulroy/status/1989037900332241004 - turns out its stupid easy to deploy @EffectTS_ HttpApi's on cloudflare https://t.co/VuBvBv9rfH - **SamNewby_** (2025-11-13): https://x.com/SamNewby_/status/1989008648916381926 - think a few of you may like this. explorer for viewing the state of your Cloudflare workers, KV, D1 and more when you're developing locally all bundled up into a little binary called wrkrscope https://t.co/EvIr9nNMrr - **fayazara** (2025-11-13): https://x.com/fayazara/status/1988919409562235381 - Want to learn how I made this webhook tester with @Cloudflare durable objects? https://t.co/cTIXkQ0eEt - **southpolesteve** (2025-11-13): https://x.com/southpolesteve/status/1988815223487361094 - Several people have asked me where we use CapnWeb at Cloudflare. Read all about it! https://t.co/tMicxjStdP - **swyx** (2025-11-13): https://x.com/swyx/status/1988815069556466049 - Huge turnout to celebrate @mikeal’s memory tonight. most of us using (and hating) Node.js today don’t know what he did to keep the community open and thriving. When I was just a replyguy in 2017 he was kind and patient with me. as aiewf grows to 6000 next year, funny to see my old interactions with him now talking about how nonlinear it is to grow these things to 600. RIP and thank you. - **MattieTK** (2025-11-13): https://x.com/MattieTK/status/1988783197803671864 - We heard you like technical deep dives 🤓 https://t.co/SpXXs7QU8O - **kimmaicutler** (2025-11-13): https://x.com/kimmaicutler/status/1988764079826313617 - @pitdesi so much Bay Area history on sale these days. you can also buy the MCM mansion of the landlord responsible for triggering rent control in SF https://t.co/XCTLQCj9rv - **marlene_zw** (2025-11-12): https://x.com/marlene_zw/status/1988689156906824114 - Made it to Demo Days #11 at the OpenAI offices in London🇬🇧✨ The awesome @mattzcarey welcoming us before the demos start🥳 https://t.co/3i1DbKszlo - **DanielLurie** (2025-11-12): https://x.com/DanielLurie/status/1988680675411779713 - This new law enforcement sobering center marks a fundamental change in San Francisco: If you do drugs on our streets, we will arrest you. And with this new resource, we will give you a real chance to enter recovery. https://t.co/iXxeFWGbie - **JeffDean** (2025-11-12): https://x.com/JeffDean/status/1988674389773357077 - Exciting expansion! @Waymo now serves the whole SF Bay Area Peninsula from SF to San Jose and is taking riders on freeways. https://t.co/fNgqQtHB7b https://t.co/m2paGRRDIO - **threepointone** (2025-11-12): https://x.com/threepointone/status/1988666951481323732 - seeing @leeb back on stage in full form, feels like 2015 all over again https://t.co/9sIQwUsvhp - **kane** (2025-11-12): https://x.com/kane/status/1988666827556417658 - The reason Waymo isn’t allowed to operate at SFO yet is bc former progressive supervisor Aaron Peskin led an effort to explicitly block it. This is why SJC already has Waymo and SFO doesn’t. https://t.co/G2eONRZnw5 - **zebassembly** (2025-11-12): https://x.com/zebassembly/status/1988654567698190785 - I hate Go so much, what an awful language to build production systems in. I love adding hours to this task because I have to refactor this code to deal with a cyclical import, I certainly don't have anything more important than dealing with an arbitrary code-style requirement! - **thomasgauvin** (2025-11-12): https://x.com/thomasgauvin/status/1988644719849599213 - Merged! This doc goes over the connection lifecycle of a database connection when using Hyperdrive. And it's really the magic at the core of Hyperdrive. Db connections require multiple round trips to set up ( 7x going from your Worker to the db to setup a connection) for TCP, TLS, and the db auth. This is the connection setup handshake Hyperdrive makes that super fast by letting you setup connections using the special Hyperdrive connection string (env.HYPERDRIVE.connectionString) within your Worker code That will actually connect you to a _Hyperdrive instance_ running right next to your Worker. That means that your db connection handshake (the 7 roundtrips) happens all next to your Worker, eliminating cross-region network latency Once connected, your driver sends the query and that happens in a single round trip across regions within Hyperdrive to get to your db One the other side, Hyperdrive pools database connections close to your database and makes it easy to reuse them across Worker invocations. So that means that creating a new db connection on every Worker invocation is super fast because connections are made to Hyperdrive, and then Hyperdrive reuses some of its open connections. This makes database access from Workers much faster and more scalable thanks to the built-in connection pooling. This is why we made is 1) super easy to use, just drop in your connection string and get a new Hyperdrive connection string and 2) available in all plans, free or paid Check out the docs! 👇 - **dillon_mulroy** (2025-11-11): https://x.com/dillon_mulroy/status/1988356454542168390 - i get asked all the time on twitch how to get started with open source. my answer: make the tools you already use better by fixing problems you have with them for everybody https://t.co/z0a3VV3Uhe - **ricklamers** (2025-11-11): https://x.com/ricklamers/status/1988304740460183739 - This is cool, GEPA/DSPy-style prompt optimization for your MCP tool prompts by @_laurynas from @fiberplane https://t.co/8lOQGlDXIC - **ritakozlov** (2025-11-11): https://x.com/ritakozlov/status/1988243560752169264 - one thing amazon got right: obsess over customers, not competition: “we are never satisfied with our offerings, and we have a peculiar culture at Amazon where we wake up every morning terrified, not of our competition, but of our customers” - bezos - **jam_croissant** (2025-11-11): https://x.com/jam_croissant/status/1988104675502035280 - Liquidity cuts ✂️ both ways… 1/ “Liquid private equity” & “private credit” funds got the green light 🚦from the SEC — after massive lobbying by an industry desperate for liquidity in a market with no buyers. ⸻ 2/ The irony❓ These same “liquid” vehicles might be the thing that undoes private markets altogether. ⸻ 3/ When crises hit, what do we always do ⁉️ 👉 We suspend mark-to-market. That’s how PE & private credit survived the last few crises — 🚫 NO forced marks, 🚫 NO runs. ⸻ 4/ But “liquid” PE & PC funds mean something new: They’ll be priced DAILY… MARKED… VISIBLE. ⸻ 5/ That means private assets — long valued by “model” not “market” — will suddenly have a market. And that market will soon speak. 🗣️ ⸻ 6/ Once the REAL marks emerge, the cascade likely begins: 🔹 Downward repricing 🔹 Margin pressure 🔹 Deleveraging in an industry built on leverage & opacity ⸻ 7/ The PE & credit giants lobbied for liquidity… They may have just lobbied for their own reckoning. 💥 Liquidity cuts ✂️ both ways… - **ritakozlov** (2025-11-10): https://x.com/ritakozlov/status/1988022779946426709 - huge fan of zed! this is so so cool https://t.co/kXsyx5hVTm https://t.co/qSvvr6KyNY - **zebassembly** (2025-11-10): https://x.com/zebassembly/status/1988007506400907591 - Very cool to see Zed build on Cloudflare! I've been trying Zed for the last week and this so close to replacing Cursor as my go-to editor, absolutely love the vision of this team. - **advaithj1** (2025-11-10): https://x.com/advaithj1/status/1988003886037414320 - Cloudflare Workers users: we have updated our rate limit system so you should no longer get incorrect 1015 rate limit errors while accessing Discord's API! I'm glad we've finally addressed this pain point to unlock reliable Discord integrations on Workers - **zeddotdev** (2025-11-10): https://x.com/zeddotdev/status/1987957296622870909 - "Our current backend has been chugging along since basically the beginning of the company. We needed a full reboot of our backend... Enter Zed Cloud." Learn about how we're rebuilding our cloud infra in Rust (obviously) and Cloudflare: https://t.co/g6p7X1pum6 - **destraynor** (2025-11-10): https://x.com/destraynor/status/1987936013524549986 - For years @patrickc + @collision told me "You should add @link, it'll increase your conversion rate" I always figured Link was only useful for consumer-y impulse buys, not Serious Business Stuff™️ But...TL;DR: we added @link and it increased our conversion rate. Lesson learned https://t.co/1rZpt4Oi1v - **boristane** (2025-11-10): https://x.com/boristane/status/1987899431400665429 - landing in sf in 7 hours I'm in town for the week, still a few empty spots on the calendar hit me up, dms open! https://t.co/KLPn6cjuIO - **jsneedles** (2025-11-10): https://x.com/jsneedles/status/1987870046358581619 - Another cool aspect of this architecture: We tolerate failures super well. Obviously the ingest layer is fallible, but also the simplest. Once we have a file on R2, we can re-ingest it, re-queue it, etc. Also, if for whatever reason we have errors in parsing, a separate pipeline exists (that basically mirrors this one) just to track errors with the raw payload and some client info. Thankfully it’s very dormant! Anytime we fail to ingest a file into CH that goes through a large number of retries (with back off) and eventually can land in a DLQ. This can be replayed through the original pipeline. We also (try to) insert info about that failure into a dedicated CH table for error tracking. - **yusukebe** (2025-11-10): https://x.com/yusukebe/status/1987724706934583807 - 17 lines file uploader with authentication with Hono and Cloudflare Workers. Beautiful✨ https://t.co/OKqBWRFOn3 - **DanielleFong** (2025-11-09): https://x.com/DanielleFong/status/1987639964042232256 - once again people must read @paulg https://t.co/Tlt4iBLynH https://t.co/uZjXcUWztY https://t.co/cw91xnA2kd - **Markzandi** (2025-11-09): https://x.com/Markzandi/status/1987549339284045904 - It’s hard to overstate the significance of the soaring stock prices of artificial intelligence companies to the economy. Spending by well-off Americans, driven by their surging stock portfolios, is the single most significant driver of growth. This so-called wealth effect is responsible for nearly half a percentage point of real GDP growth over the past year, accounting for one-fourth of the economy’s overall growth. The wealth effect has been a tailwind to growth, more or less, since recovering from the Global Financial Crisis. Of course, housing wealth is also up a lot, and the well-to-do are still holding onto a big pile of cash that they accumulated during the pandemic shutdowns. But given the outsize gains in wealth compared with the growth in incomes and the economy more broadly – household net worth is now eight times after-tax income, compared with an average of 5.5 times in the decades between WWII and the GFC – it is prudent to consider how sustainable this is. And if it isn’t, what does it mean for future growth? - **kentcdodds** (2025-11-09): https://x.com/kentcdodds/status/1987526330938958278 - @theo @dsp_ Here, now you've seen a months old implementation that is not a hardcoded API key: https://t.co/pNKSJaJNCK (the OAuth authorization server is here: https://t.co/sXpe5sCU4k) - **pmarca** (2025-11-09): https://x.com/pmarca/status/1987353696129413307 - 👀 https://t.co/2Kk3LRUgxq - **simonw** (2025-11-09): https://x.com/simonw/status/1987329464695726197 - TIL that MCP uses the little-known Dynamic Client Registration feature of OAuth, the first time that feature has been used by widely deployed software https://t.co/40WCQYgJSn https://t.co/pTmpBIaMwr - **KentonVarda** (2025-11-08): https://x.com/KentonVarda/status/1987208904724652273 - The big advantage of MCP over OpenAPI is that it is very clear about auth. OpenAPI supports too many different auth mechanisms, and the schema doesn't necessarily have enough information for a robot to be able to complete the auth flow. https://t.co/RJP6nEpPMs - **thdxr** (2025-11-08): https://x.com/thdxr/status/1987202620394774685 - a truth people hate and really want to believe otherwise is if you get big enough you actually can figure out how to make money later people remember the obviously stupid attempts at this and don't notice all the ones that worked - **grinich** (2025-11-08): https://x.com/grinich/status/1987197567592276316 - feels like the era of hype is over. signups, views, free users, github stars.. all meaningless vanity metrics what matters now is revenue, margin, retention, and the sound of cash hitting your bank account the tide is changing. and we're about to see who’s been swimming naked - **threepointone** (2025-11-08): https://x.com/threepointone/status/1987195553201979693 - sandboxing from the future ™ - what if eval had superpowers - make a worker on the fly from strings - instant start/stop - control what’s available to call inside it - control all outgoing i/o here’s an example: https://t.co/FtzPn4c1M6 https://t.co/VmvOK1z0Cw - **dexhorthy** (2025-11-08): https://x.com/dexhorthy/status/1987145903304499401 - advanced context engineering - pep turned me onto the idea of “context shards” and I love this framing - big fan of designing workflows to be incremental and resumable - great way to keep context usage efficient https://t.co/krSLT3TaEx - **tobi** (2025-11-08): https://x.com/tobi/status/1986997846428574063 - very important in companies. Keep the ones willing to "truth nuke" close. They are load bearing. https://t.co/T3UK1faab4 - **acoyfellow** (2025-11-08): https://x.com/acoyfellow/status/1986970161912455359 - Is there some kind of Cloudflare durable objects fan community? If not we should make one - **ArtemisConsort** (2025-11-08): https://x.com/ArtemisConsort/status/1986963360471261349 - The reason it’s incredibly hard to break out of a “kind lie” doom loop is because the longer it’s gone on, the more unkindness is needed to get back on track. And the fact that you’re in this situation to begin with probably means your culture will hate that. https://t.co/vaPTX2rkvo https://t.co/t9jajvdeg3 - **clairevo** (2025-11-07): https://x.com/clairevo/status/1986943423358869625 - Can you believe it's been a year since I murdered product management on stage at the @lennysan summit? Time flies when you're collapsing the talent stack. https://t.co/oNsOHow9eB - **dinasaur_404** (2025-11-07): https://x.com/dinasaur_404/status/1986862956920062122 - We're looking for engineers to join the Workers Deploy & Config team, responsible for: - the system that deploys every Cloudflare Worker - Workers for Platforms - Preview URLs, Versions, Gradual Deployments - and more! We have the best EM & engineers on the team, come join! - **morganlinton** (2025-11-07): https://x.com/morganlinton/status/1986808469312417878 - I think this article from @clairevo is pretty incredible. She predicted back in October of last year what has become more true by the week in 2025. If you somehow haven't read this yet, you should. Founders, engineers, and of course PMs, there's nuggets to everyone in this. Link in first comment below. And no, I'm not being paid to share this, I just think this article is🔥 - **threepointone** (2025-11-07): https://x.com/threepointone/status/1986778049472315837 - - Sync Conf all of 12th https://t.co/XNn0hroo5I - This weird secret CF thing on 13th morning https://t.co/QGcxdKhJZ0 - This ai agents meetup with friends from anthropic/openai 13th evening https://t.co/3kUBKwBGy7 https://t.co/2kAPPf6ZWn - **dillon_mulroy** (2025-11-06): https://x.com/dillon_mulroy/status/1986568399149760635 - i hope yall are ready for me to be insufferable on the tl about durable objects - **dillon_mulroy** (2025-11-06): https://x.com/dillon_mulroy/status/1986541931850309744 - @igalklebanov @Cloudflare no beef, i left vercel on good terms and proud of everything we shipped. i’ll still be cheering on all the incredible work my friends and former colleagues are doing there - **threepointone** (2025-11-06): https://x.com/threepointone/status/1986533883253338140 - OH MY GOD WHAT 🤩 https://t.co/vcDccODGae - **dillon_mulroy** (2025-11-06): https://x.com/dillon_mulroy/status/1986529363391627607 - life update... very excited to share i'll be joining @cloudflare to work on the developer platform and workers! the momentum and incredible work that the team has shipped over the past few months has been inspiring and exciting to watch. stoked to get to be a part of it https://t.co/wa2mEFPUwF - **topmass** (2025-11-06): https://x.com/topmass/status/1986509075044319541 - tanstack start running fully on workers: wrangler init thats it! 🎷 https://t.co/i9VwSCPLX2 - **southpolesteve** (2025-11-06): https://x.com/southpolesteve/status/1986507503312551995 - @thdxr nooo. Is it something about batching? Can I interest you in https://t.co/yDrYvKTtqx instead? - **martindonadieu** (2025-11-06): https://x.com/martindonadieu/status/1986472376884719884 - Memory and CPU cause and inspect. Currently I have no clue why it cost me this CPU time. It would be nice to have a top 10 of functions that cost the most CPU time, like we have in D1 for queries. Be able to set a threshold for alerts, like error rate, etc and I get emails. Same for Client disconnected, and Exceeded Memory, Exceeded CPU Time Limits, Uncaught Exception: I want the top 10 causes of it. Right now I know I have them but have no clue how to find them. I switched all pages to workers, as it says in many places it's better, but lost all observability and stats; it says now 0 in metrics everywhere. Cannot load balance 2 different workers. ( like i have one with Smart Placement and hyperdrive and one without and D1 read replicate and i need another worker to compare them. Cannot click in the binding Analytics engine to get to it. - **jeresig** (2025-11-06): https://x.com/jeresig/status/1986471502716211501 - @dinasaur_404 @Cloudflare Skew protection would be incredible! I feel like it would require some level of integration with whatever frameworks are built on top of it - but being able to ensure that backend APIs can stay intact across deploys would be awesome. - **mdhruvill** (2025-11-06): https://x.com/mdhruvill/status/1986468202801733915 - I don’t think there’s currently a way to separate build variables and secrets for production and preview builds. Right now, I have to run a prebuild script that checks the branch from an environment variable and sets the right values based on the environment. It would be to have this feature. - **Matthewtylerayl** (2025-11-06): https://x.com/Matthewtylerayl/status/1986468100376842357 - @dinasaur_404 @Cloudflare Monorepo support?😅 I'd love to move my builds into CloudFlare but having it rebuild individual workers is a little hard - **dinasaur_404** (2025-11-06): https://x.com/dinasaur_404/status/1986460769576206632 - morning yall! 👋 we're working on improvements around @Cloudflare Workers versioning/deployments. if you have any feature requests around this, let me know! can be anything from version skew protection, improvements to gradual deployments, UI requests, and more! - **ritakozlov** (2025-11-06): https://x.com/ritakozlov/status/1986447956526985249 - ICYMI: you can now connect to applications running in our VPC directly from your cloudflare worker full details 👇 https://t.co/3PO15onreQ - **threepointone** (2025-11-06): https://x.com/threepointone/status/1986370692858663261 - this is really weird SF folks sign up asap https://t.co/VEBcpEXyYk - **_chenglou** (2025-11-06): https://x.com/_chenglou/status/1986289051226546448 - Some thoughts: - Semantic search is nothing new, but the little innovation here is on the data collection side. Cursor agents have metrics for success & failures (e.g. user accepts/rejects), and we can look into those sessions’ searches, and optimize the embeddings so that next release, agents can just directly search better and better - Grep + semantic search is also pretty established; real-world search has always been hybrids (tf-idf and bm25 as pre/post-filter + semantic search) - The real unlock here, in this democratized AI era, is that you can over-include false positive results and have AI trim things down, rather than under-include results (bad) just because they’re too noisy for human reading. Over-including results is basically the premise of e.g. Exa and other AI search engines - “Semantic” is pretty overloaded nowadays; maybe a better term would be “similarity search”. But the ship has sailed - **samgoodwin89** (2025-11-06): https://x.com/samgoodwin89/status/1986260824579883214 - It means we can have a simple compiler plugin to infer least-privilege policies. Just write and run your business logic. https://t.co/uHjNFmrlqx https://t.co/kPOTjPY1oP - **acoyfellow** (2025-11-06): https://x.com/acoyfellow/status/1986233008999776581 - @sudbalaji @irvinebroque @AnthropicAI i'm experimenting with an idea around abstracting cloudflares infra into a tool call.. might be of interest https://t.co/eTRRrRopmL - **sama** (2025-11-05): https://x.com/sama/status/1986204243997688184 - A thing often in common among great startup investors, founders, and researchers: Trading making a lot of small mistakes in exchange for getting a few giant wins. (Surprisingly many people seem to prefer a few big mistakes in exchange for a lot of small wins.) - **craigsdennis** (2025-11-05): https://x.com/craigsdennis/status/1986194044327018867 - Hey @threepointone , you're gonna be in SF next week, right? We should totally go to this. Any idea what it is? https://t.co/TF4zwYWudB - **dcarter_js** (2025-11-05): https://x.com/dcarter_js/status/1986170031479091510 - to write clearly is to think clearly. all of the most effective and smartest people i’ve ever worked with were/are very very good at writing - **swyx** (2025-11-05): https://x.com/swyx/status/1986165502897758491 - i truly do not care what koding agent you use, but if you dont have a slackbot that you can just @ and it converts your slack threads into PRs you are living in the caveman era and actively not even giving AI a chance out of spite https://t.co/HTopiNzpoN https://t.co/6F23u1wJnE - **dsp_** (2025-11-05): https://x.com/dsp_/status/1986162218354766273 - We (MCP Core Maintainers) accepted a proposal to add a long running Tasks concept to the next version of MCP: https://t.co/ehzMNU3rnW . There are details to iterate on, but the direction is accepted. - **acoyfellow** (2025-11-05): https://x.com/acoyfellow/status/1986120889025343672 - exactly this. ai writes code better than running tools. I started building anytool to answer "where should this code run?": - generate code on demand (ai writes tools) - execute in isolated containers (@bunjavascript) - full npm support + security boundaries - editable, cached & reusable (don't regenerate) early version but code generation + safe execution feels like the future - **pierceboggan** (2025-11-05): https://x.com/pierceboggan/status/1986116693819859024 - It is clear developers need a unified experience for all coding agents. We introduced a new VS Code primitive view to address this: Agent sessions. From one view, kick off, monitor, and review all of your coding agents in @code - local or remote, from Copilot or agents like Codex. https://t.co/8pbQsi4ht6 - **KentonVarda** (2025-11-05): https://x.com/KentonVarda/status/1986116489431159266 - So Anthropic agrees with us: AI agents are better at writing code than running tools, and future agents will probably do everything by writing code. But where should this code *run*? My recent Cloudflare Connect talk was all about exactly this! https://t.co/dzJuzBIA3J - **elithrar** (2025-11-05): https://x.com/elithrar/status/1986079597692121329 - this is a big one. build on the Workers stack + call into the services & databases you have in your existing cloud w/o having to punch holes in your VPC firewall. https://t.co/UZjvIh8j3O - **thomasgauvin** (2025-11-05): https://x.com/thomasgauvin/status/1986078650249175480 - We just released our first milestone of our vision for a global vpc - Workers VPC Services! Now, you can access your external resources in external private networks from Workers, accessible via bindings for built-in security and simpler DX (deep dive in the blog 👇) https://t.co/59IfJjJKwS - **nevikashah** (2025-11-05): https://x.com/nevikashah/status/1986072889116721623 - This is truly one of the best ELI5s of Durable Objects I’ve read. If DOs have ever felt confusing, @boristane broke👏🏽it👏🏽down with analogies, diagrams, and code snippets to make it all click ✨ https://t.co/Gp7dBXRbi7 - **arpit_bhayani** (2025-11-05): https://x.com/arpit_bhayani/status/1986070799498051873 - Cloudflare shared some data on Internet connections at scale, and the numbers are really interesting... Half of all TCP connections consist of just 12 packets or fewer. The median response from Cloudflare servers is only 4.8 KB for HTTP/1.X and 6 KB for HTTP/2. Most web interactions are incredibly tiny. But here's where it gets interesting. While the median connection lasts just 4.7 seconds, some connections stay alive for days. The median connection sends 3.78 times more data than it receives, showing just how download-heavy the modern web is. The distribution follows the classic 'elephants and mice' pattern. A small fraction of connections carry millions of packets (video streams, large downloads), while the vast majority remain lightweight (API calls, small web objects). Most connections carry just one HTTP request, even with HTTP/2's multiplexing capabilities. This holds true for both user-driven traffic and automated datacenter traffic. The Internet is far more 'bursty' and transient than we might think. Hope you found this interesting. - **boristane** (2025-11-05): https://x.com/boristane/status/1986068986283356189 - the question I get the most at every conference or meetup: "what even are durable objects?" I spent the past 2 years building almost exclusively with durable objects, and today I'm compiling my notes and favourite patterns into a single blog post https://t.co/fyYn1Daeb7 https://t.co/dJY6FRetrW - **MattieTK** (2025-11-05): https://x.com/MattieTK/status/1986060768258003448 - 👋 Cloudflare peeps! How do you manage with inspecting local resources like KV/R2/D1/DOs? Have you ever encountered a challenge where you wanted to know more about your current resource state? Do you use any tools? Is this just not a problem for your case? - **samgoodwin89** (2025-11-05): https://x.com/samgoodwin89/status/1985952459110715842 - @irvinebroque I was thinking of plugging in to the effect language service plugin to interpret these for users. They already do similar things for simplifying error reporting. - **boristane** (2025-11-05): https://x.com/boristane/status/1985930947385311546 - anthropic blog: https://t.co/BdPHMReRL3 - **boristane** (2025-11-05): https://x.com/boristane/status/1985930942850924801 - anthropic just published a blog highlighting a pattern I've been using for weeks: a search tools tool instead of loading all tools into the context window of the model, I load a "search tools" tool the model can use to browse a catalog of tools just-in-time tool discovery https://t.co/EFSm1k2KfB - **boristane** (2025-11-05): https://x.com/boristane/status/1985930945686392929 - the anthropic blog focuses on how this could be used within a filesystem and paired with mcp and sandboxed code excecution I like to keep things much simpler, this runs in a @cloudflaredev durable object, I'm yet to try the new dynamic worker loader https://t.co/90j0pO25N2 - **ethanniser** (2025-11-05): https://x.com/ethanniser/status/1985895994463306005 - You can just speak at meetups https://t.co/kJVydMezkD - **MarcSelwan** (2025-11-05): https://x.com/MarcSelwan/status/1985883673636614429 - Hey friends - we’ve been working on a UI for R2 Data Catalog - what would you like to see? Are there any data lake/data engineering centric UIs you really like? Honestly, I haven’t really found one I’m passionate about. (Hence why I worked on https://t.co/sW8cJtlLAJ to experiment with ideas) - **simonw** (2025-11-04): https://x.com/simonw/status/1985852549317140986 - Interesting proposal here from Anthropic to convert MCP tools into code functions that your coding agent can then discover and execute after reading the file system, albeit with a huge "draw the rest of the owl" factor in that implementation is left as an exercise for the reader! https://t.co/x5uAnW9H2H - **AnthropicAI** (2025-11-04): https://x.com/AnthropicAI/status/1985846791842250860 - New on the Anthropic Engineering blog: tips on how to build more efficient agents that handle more tools while using fewer tokens. Code execution with the Model Context Protocol (MCP): https://t.co/PeStmufIkp - **astuyve** (2025-11-04): https://x.com/astuyve/status/1985831530153800106 - Cold starts don't just impact serverless functions. All engineers should understand how to profile and improve their application cold boot time. It's on the critical path for scaling up during an outage, it's key to faster deployments, and the general dev experience https://t.co/HxCcSPuasX - **mwylde** (2025-11-04): https://x.com/mwylde/status/1985823717184979277 - This is Thursday! https://t.co/zXu9oIMWUa - **samgoodwin89** (2025-11-04): https://x.com/samgoodwin89/status/1985818898986647828 - Here's a short video walking through how alchemy-effect guarantees least-privilege IAM policies for DynamoDB. https://t.co/rsTEVNZusT - **LakeAustinBlvd** (2025-11-04): https://x.com/LakeAustinBlvd/status/1985767892835860572 - I am hiring for a Product Manager of Cloudflare Radar. DMs open! Radar is where we share our view of the Internet with the world, and we have so much more to build and so many more stories to tell. If that challenge sounds exciting, please reach out! https://t.co/MaFamhLrL8 - **BraydenWilmoth** (2025-11-04): https://x.com/BraydenWilmoth/status/1985730757155803455 - 60+ pieces of feedback I've noted down so far from your responses. User feedback → Notes → Jira. A lot of good suggestions and quick wins. Some a little out of scope of dashboard but heard and sharing nonetheless! ✍️ Keep 'em coming. https://t.co/Q5SmpzjqML https://t.co/5QC6EBKVot - **stewartbrand** (2025-11-04): https://x.com/stewartbrand/status/1985539320070164732 - @tentcontrol @stefanoscalia Psychedelic drugs. Personal computers. Great music. Avant garde going pop. The predecessors to Burning Man. Do it yourself. - **stewartbrand** (2025-11-04): https://x.com/stewartbrand/status/1985506564867080344 - @stefanoscalia Mistakes continue to chafe. Successes disappear into the normal. - **destraynor** (2025-11-03): https://x.com/destraynor/status/1985488484711174181 - @patrickc Agree: selling to European companies is such a heavier lift: data residency rules, localisation demands on everything. endless contractual adjustments, zero risk tolerance in the larger companies when procuring. It’s one of the less discussed EU challenges (vs funding etc) - **tjcages** (2025-11-03): https://x.com/tjcages/status/1985486703964872995 - @patrickc very interesting. more insights like this please - **patrickc** (2025-11-03): https://x.com/patrickc/status/1985468907747172552 - An interesting trend we're noticing at Stripe: US startups are pulling ahead of their peers elsewhere. These charts show averaged revenue growth for software startups in each location. US startups typically grow somewhat faster than those elsewhere. However, since mid-2023, US companies have accelerated a lot. Interestingly, this is not just because of AI startups: if we strip those out, there's still a big divergence. Our leading hypothesis is that US startups (even those that aren't AI companies as such) are adopting new technologies (AI, stablecoins, etc.) faster than companies elsewhere. (This pattern of faster adoption among US companies was also seen with the internet itself.) Whatever the cause, the pattern is striking. [Methodological note: this pattern appears to hold beyond Europe as well.] - **cakemakerjake** (2025-11-03): https://x.com/cakemakerjake/status/1985393940971823127 - @BraydenWilmoth Enable/disable DNS records - **grinich** (2025-11-03): https://x.com/grinich/status/1985390515093901359 - Last week we shipped a massive new @WorkOS feature: 🔑 API keys 🔑 This is a fully managed system for providing access to developers and agents. Comes with drop-in UI via widget, and works seamlessly with existing orgs and permissions. Demo from @NickyHolden https://t.co/eBNu7eIuA8 - **burakkarakann** (2025-11-02): https://x.com/burakkarakann/status/1984963118900949250 - We are hosting our first-ever meetup in SF, together with @Cloudflare! I will speak about bringing AI into data pipelines, and @mwylde from Cloudflare will talk about how they built their new R2 SQL engine. https://t.co/wj9E7aiU7G - **BillSimmons** (2025-11-02): https://x.com/BillSimmons/status/1984832449663418400 - Yamamoto pitching tonight and getting key outs is inconceivable. - **jaredhanson** (2025-11-02): https://x.com/jaredhanson/status/1984828219900772606 - @zeeg You can’t watch that 9th inning and *not* like baseball! - **CaitlinClark22** (2025-11-02): https://x.com/CaitlinClark22/status/1984828031811113236 - This is the best baseball game I’ve ever seen - **msuster** (2025-11-02): https://x.com/msuster/status/1984827963033206979 - OMG OMG OMG OMG What a game - **BillSimmons** (2025-11-02): https://x.com/BillSimmons/status/1984827145500442637 - Holy fuckballs this game is incredible! - **threepointone** (2025-11-01): https://x.com/threepointone/status/1984596776758784330 - local-first people: what's the thinking on storing sensitive user data in localStorage/IndexedDB; what's the fallback/workarounds you're thinking of. how do you implement TTL, all that. are there special browser apis for this usecase I'm not aware of. tagging @aboodman @kylemathews @schickling - **nandafyi** (2025-10-31): https://x.com/nandafyi/status/1984327860644168042 - split deployments https://t.co/H1bTyV1iZu - **yagiznizipli** (2025-10-31): https://x.com/yagiznizipli/status/1984277662140174730 - improving @cloudflare workers text encoder encode method by 6x https://t.co/EUt3hb92c2 - **yomnashousha** (2025-10-31): https://x.com/yomnashousha/status/1984253537375920345 - new preview button in the build details page to access your preview url :) https://t.co/F8byBt4daj - **hichaelmart** (2025-10-31): https://x.com/hichaelmart/status/1984091211637170605 - Wait wait, I meant a 91% boost https://t.co/wIXbqOhd9x https://t.co/GFXqAroKnM - **yagiznizipli** (2025-10-31): https://x.com/yagiznizipli/status/1984053694548373540 - i realized something today. deeper i go as a system engineer, the more i am excited. but i don’t know why… - **thomas_ankcorn** (2025-10-30): https://x.com/thomas_ankcorn/status/1984010927541567987 - What if observability could be pretty too? https://t.co/B7kPRM9ZOM - **juliannaelamb** (2025-10-30): https://x.com/juliannaelamb/status/1983990778549666144 - We started Stytch 5.5 years ago to build the next generation of identity infrastructure. We’ve come far in that time, but there’s still so much more to do as the way we interact with applications and the threats to identity security are evolving rapidly with AI. Today, I’m excited to share that we’re joining forces with Twilio to accelerate our roadmap by bringing identity and communication infrastructure together to help you build a new generation of AI-driven applications. I vividly remember sitting in the backyard of my mom’s house one Saturday amidst the 2020 lockdowns, working on what would become the Stytch API. One of the very first pieces of code that I wrote was a Twilio integration for doing SMS OTP login. It was the first end to end auth product that I built as part of the initial POC for Stytch. I remember sending my mom an OTP code and running in the house to make her finish the login, one of the first end to end Stytch authentications. The next product I worked on was email magic links, and as I was going through the setup, I excitedly sent @McStempel the onboarding flow for SendGrid (by then a part of Twilio), impressed with how thoughtful their developer experience was. Twilio has been a cornerstone of the Stytch product from before it was even a product, when it was mostly an idea and a little Go server running on my laptop. So it feels incredibly full circle to be joining forces with Twilio to double down on building identity infrastructure that makes securing applications simple for developers. Building Stytch has been an amazing journey, I’ve gotten to work with incredible, customer obsessed people, been challenged in ways I could never have imagined, and I’m so proud of the work we’ve done. But we’re just getting started. I'm excited for this next phase of the journey, to build even better products at a scale that we couldn’t even dream of when those first OTP codes were sent. - **thomas_ankcorn** (2025-10-30): https://x.com/thomas_ankcorn/status/1983990188562165993 - @nandafyi Low key I suspect the days of the figma designer as a job are numbered 🫠😬 - **tannerlinsley** (2025-10-30): https://x.com/tannerlinsley/status/1983982682456674429 - Heyo! ⛅ @Cloudflare has raised the stakes on the @Tan_Stack Hackathon to a total of $140k in total prize value! 🔥 I *really* can't wait to see what people build now. 🔗🧵↓ https://t.co/IuS3AGgLrR - **mwylde** (2025-10-30): https://x.com/mwylde/status/1983931464388383218 - Congrats @akshaykagrawal and the whole Marimo team! https://t.co/XWx92u2O5g - **yomnashousha** (2025-10-30): https://x.com/yomnashousha/status/1983901904217284723 - Couldn't resist joining in on all of the Cloudflare performance improvement fun Here are 3x faster build starts :) Side effect of Builds built on Workers --> faster container startups https://t.co/Sfqfx57IHg - **ritakozlov** (2025-10-30): https://x.com/ritakozlov/status/1983895771242996198 - love to see it 🔥 h/t @boristane https://t.co/SfVuuzDlHr - **southpolesteve** (2025-10-30): https://x.com/southpolesteve/status/1983698770828390805 - @GameRoManDev @jarredsumner We can absolutely help make that happen - **jarredsumner** (2025-10-29): https://x.com/jarredsumner/status/1983653358612492739 - what if `bun build --target=cloudflare` https://t.co/Taq34kiyrM - **RobbieSap90** (2025-10-29): https://x.com/RobbieSap90/status/1983615916383281456 - Did you know Workers Builds are built on top of Containers? We're able to build the best product for developers when we build with that product. After some hard work behind the scenes, the time it takes to start your build should be even faster — by 3.3x in some cases. - **thdxr** (2025-10-29): https://x.com/thdxr/status/1983602942830899230 - the concept of 0 is crazy, so hard to really hold the concept in your head and visualize it accurately but if you add up everything anime pfp accounts have ever shipped in total you get pretty close - **mwylde** (2025-10-29): https://x.com/mwylde/status/1983600263991861639 - If you're in NYC, come hear @aninibread talk about R2 SQL, our new distributed query engine next Monday! https://t.co/UbpGKXgByT - **zebassembly** (2025-10-29): https://x.com/zebassembly/status/1983563664234827789 - I recently finished a project to move the @Cloudflare dashboard to @rspack_dev with it finally hitting production today. The performance gains have been incredible, 30x faster builds and 60x faster HMR times. It's made our lives so much easier, several engineers have come up to me thinking the dev server was broken because it was so fast! But my favorite feedback was this reply I just got where the faster dev experience made this fellow engineer's day less frustrating by removing the need to wait around for webpack. If the webpack codebase you're maintaining is suffering from performance issues, please give @rspack_dev a try! For the ~20 hours I spent on this project I expect we'll save 1000 hours across the entire company over the next year. - **threepointone** (2025-10-29): https://x.com/threepointone/status/1983543626073567622 - chatgpt oauth is probably going to be a massive breakout moment - **ritakozlov** (2025-10-29): https://x.com/ritakozlov/status/1983537068182749426 - if you're not vibe coding prototypes as a PM, you're ngmi - **hichaelmart** (2025-10-29): https://x.com/hichaelmart/status/1983450613309755730 - @jarredsumner But in reality I think you’re much better off measuring more realistic (larger) workloads, and comparing on a serverless platform, which is what CF Workers is. I still think Bun will be faster! But more like the results Vercel published. https://t.co/2CwCthKjY7 - **hichaelmart** (2025-10-29): https://x.com/hichaelmart/status/1983441161021993069 - @jarredsumner I mean. https://t.co/UpZSvauwYB - **garrytan** (2025-10-29): https://x.com/garrytan/status/1983411791678824673 - Most people are still running scarcity scripts from a world that’s already obsolete. What’s happening right now (in AI, in energy, in materials, in biology) is the shift from allocation to creation. But our political and cultural systems were built for managing scarcity: who gets what slice, not how to expand the pie. When those systems meet exponential tech, they panic. They see threat instead of possibility. The tragedy is psychological: if you believe the world is zero-sum, abundance looks like chaos. But if you believe intelligence and coordination can grow, abundance looks like the next stage of civilization. - **mwylde** (2025-10-29): https://x.com/mwylde/status/1983333346295066761 - As a big user of workers internally (fun fact: the developer platform is by and large built on the developer platform!) I am so excited for this https://t.co/O5w5Cypqbl - **elithrar** (2025-10-28): https://x.com/elithrar/status/1983283678198235228 - still one of my favorite new interactions in the Workers dashboard: every service + an example just a couple of clicks away. https://t.co/InmnLzkvuJ - **threepointone** (2025-10-28): https://x.com/threepointone/status/1983232707506127306 - - customer hadn't deployed their mcp servers in europe because of gdpr - mcp servers are built on durable objects which support jurisdiction - so we shipped a config that lets you specify the jurisdiction - durable objects stay winning live in agents@0.2.19 https://t.co/gURBCmO5L2 - **boristane** (2025-10-28): https://x.com/boristane/status/1983211992467816771 - I wrote a blog post detailing it all - it's a long one, but it has a demo repo you can fork and hack https://t.co/Dh8OOMmhtg - **boristane** (2025-10-28): https://x.com/boristane/status/1983211987677954384 - agent memory works best when you combine vector search, full-text search, and graph traversal but who wants to maintain a multi-tenant graph database? not me so I built a system to create unlimited on-demand graph databases with @cloudflaredev durable objects https://t.co/rAsxbJN4FU - **harshil1712** (2025-10-28): https://x.com/harshil1712/status/1983201005513912721 - If you missed it, you can run your @UseExpressJS on @CloudflareDev Workers! I have a quick tutorial for you to get started 👇 https://t.co/buabcbIVfe - **eastdakota** (2025-10-28): https://x.com/eastdakota/status/1983175502354624612 - Amazing progress that every @Cloudflare customer gets for free. https://t.co/asBBHowC8Y - **Cloudflare** (2025-10-28): https://x.com/Cloudflare/status/1983170809225945215 - Today over half of human-initiated traffic with Cloudflare is protected against harvest-now/decrypt-later with post-quantum encryption. https://t.co/IGziy5pcya - **boristane** (2025-10-28): https://x.com/boristane/status/1983163999819837591 - we just shipped zero-intrumentation, otel-native tracing for @CloudflareDev workers we worked directly in the runtime to add native instrumentation - no code changes - no bloated otel lib - zero perf degradation single command to opt in: wrangler deploy it just works https://t.co/iWNCgT9JCh https://t.co/rK1ZHK7buc - **WalshyDev** (2025-10-28): https://x.com/WalshyDev/status/1983163394732757139 - Remember me talking about enabling a long-awaited feature and it providing crazy good value already for our small test group (https://t.co/TpoAfWJebJ)? Well now it's live! Automatic Workers Tracing is available to everyone! Improve your apps performance and find bugs easier! https://t.co/PTGJNluYWa - **craigsdennis** (2025-10-28): https://x.com/craigsdennis/status/1983162697601720525 - Yo, did you see that @AnthropicAI let's you write your own Agents and gives you access to all the tools, just like @claudeai Code? Here's a taste of the full tutorial where I show how to get an Agent hosted on a @CloudflareDev Sandbox. It uses those new skills too! https://t.co/v7A7fS9Koe - **nevikashah** (2025-10-28): https://x.com/nevikashah/status/1983161078499708986 - huge shoutout to the incredible wizards who made this happen @zebassembly @boristane @thomas_ankcorn @JeremyMorrell and so many more ✨ - **zebassembly** (2025-10-28): https://x.com/zebassembly/status/1983159978803277883 - This has been what I've been working on for nearly a year now, super excited to finally do the first release of Automatic Tracing. Starting today anyone with Observability enabled will automatically get traces generated for their Workers that are present in the Cloudflare dashboard and can be sent to third party observability providers. We've had internal teams and select customers using this for a while and they've been able to make good performance gains through the increased understanding of what's happening in their Worker. But, today's release is only the first of many. There are several things that we didn't have time to include in this first release like context propagation for distributed tracing, user-created spans, even span events, or even some CF bindings we didn't get to instrument. All of these gaps are on the tracing roadmap and will be released as they're ready, but in the meantime there's still a lot of useful data for those that have Workers in production. - **nevikashah** (2025-10-28): https://x.com/nevikashah/status/1983159821110259990 - Ready to go? Read about it ➡️ https://t.co/FXHehkEG74 Get started ➡️ https://t.co/lMJTUze4R7 Join the chatter ➡️ https://t.co/jZ0r5BtusF - **nevikashah** (2025-10-28): https://x.com/nevikashah/status/1983159194917449805 - Don’t want to leave your existing observability tooling? 👀 Don’t worry, we got you. 🚀Configure a destination with an OTel endpoint and start exporting traces and(!!) logs to your favorite provider. https://t.co/QIH3lQXabd - **ritakozlov** (2025-10-28): https://x.com/ritakozlov/status/1983159086058463300 - otel traces are now available in dashboard and for export in cloudflare workers! go forth and debug 💪 huge congrats to the team on what’s a massive step forward for workers observability! https://t.co/APW4JLDKk6 https://t.co/hwARm0qq3d - **nevikashah** (2025-10-28): https://x.com/nevikashah/status/1983158895188103489 - Ever wondered if a long running API call is a problem across many requests? 🧐 With every span, you’ll get a relevant attributes for that operation. For example fetch() includes attributes like URL, status and duration which you can query using the Workers Observability “investigate” tab 🕵🏽‍♀️ - **nevikashah** (2025-10-28): https://x.com/nevikashah/status/1983158492379918413 - Automatic tracing? Yes, really 😎 We’ve instrumented our Workers runtime directly to trace operations like binding calls, fetch calls and handlers. No hacky wrapper code. No SDKs. 🚀Just enable and redeploy: https://t.co/jUOB6WAKol - **nevikashah** (2025-10-28): https://x.com/nevikashah/status/1983158227933298843 - 👏Today is the day👏 Workers ✨automatic✨ tracing is now in open beta! ✅Enable in seconds – no code changes required 🔎View and query trace data directly in the Cloudflare dashboard 📦Export traces (and logs!) to any external destination with an OTel endpoint https://t.co/W36ft8jbKq - **Cloudflare** (2025-10-28): https://x.com/Cloudflare/status/1983158061658190056 - Cloudflare Workers' support for automatic tracing is now in open beta! Export traces to any OpenTelemetry-compatible provider for deeper application observability, no code changes required. https://t.co/jbNZe8bNI1 - **thomas_ankcorn** (2025-10-28): https://x.com/thomas_ankcorn/status/1983157464750280816 - We shipped! https://t.co/EqWCy0F7S8 This has been so much fun to work on. - **boristane** (2025-10-28): https://x.com/boristane/status/1983157101909377045 - what is this? https://t.co/VpDqyzo5WR - **zeeg** (2025-10-28): https://x.com/zeeg/status/1983012006790918527 - The test here for many of us is easy: name three public company CTOs. I might be able to do it but it used to be a lot easier… People who are in the weeds at that scale, that you can respect, they’ve thinned out. https://t.co/SvezkbHAhT - **ccheever** (2025-10-28): https://x.com/ccheever/status/1983009149345800316 - 📈 https://t.co/M8CjGNFIap - **dok2001** (2025-10-28): https://x.com/dok2001/status/1983005313466019854 - In 2013, @agentzh showed me flame graphs for visualizing system resource usage. We got hooked on writing systemtap scripts, optimizing everything (on and off CPU), and turning it into a game: find an issue, capture a flame graph, nerd snipe, repeat. https://t.co/Cka0jytRCf - **martin_casado** (2025-10-28): https://x.com/martin_casado/status/1982967919685353564 - I'm calling it. @dok2001 is the top CTO for a public infra company at scale right now. - **MattieTK** (2025-10-28): https://x.com/MattieTK/status/1982963805417079261 - New from wrangler, no more complaints for missing IDs in your config file. 🆕 Now we generate D1, KV, and R2 resources automatically for you when you don't specify them on deployment. All kudos to @samuelpmacleod who was working on this well before I joined. https://t.co/vJ3y7kWQ75 - **grinich** (2025-10-27): https://x.com/grinich/status/1982877806095958234 - WorkOS: The Enterprise Stack for the AI Era AI companies are just B2B SaaS with a new engine. They monetize like SaaS, sell like SaaS, and scale into the enterprise like SaaS. The difference is velocity. These new products grow so fast that the old playbook of “PLG for years, enterprise later” simply does not work anymore. PMF is no longer enough. Winning your market requires crossing the Enterprise Chasm almost immediately. Enterprise auth, provisioning, RBAC, compliance, billing, IT integrations. If you wait, someone else takes your market while you are still wiring SCIM. This is why @WorkOS exists. We give developers the infrastructure they need to scale up-market on day one. Even if you have never touched WorkOS, you have already used it through products like ChatGPT, Perplexity, Cursor and many others. Today we operate 81M+ enterprise user accounts, 67M API calls per day, and 38K+ connected enterprise environments. Modern AI tools already run through WorkOS. At ERC, we unveiled six major launches expanding that foundation for the AI era: • AuthKit for ChatGPT Apps: secure OAuth + MCP so developers can connect real enterprise data to 700M+ weekly ChatGPT users. • AuthKit for Platforms: embed full enterprise identity into frameworks like Supabase and Convex for zero-friction onboarding. • Stripe Usage Sync: actual per-seat billing with no glue code. Data stays consistent and invoices are always correct. • WorkOS Pipes: the fastest and most secure way to ship integrations (Salesforce, Slack, Intercom and more). • Agent-Ready API Keys: scoped, revocable keys designed for both developers and autonomous agents. • WorkOS Studio: vibe coding for the enterprise. A collaborative AI app builder with SSO, SCIM, RBAC, audit logs, workflows and third-party connectors built-in. Internal software at the speed of a prompt. AI has already changed how software gets created. The next shift is changing what software becomes. We are still “filming the play” like early cinema: using new technology to recreate old patterns. The opportunity ahead is to invent entirely new applications that only AI-native development makes possible. WorkOS is building the enterprise infrastructure that lets teams design the next era of software itself. Enterprise-ready AI is where the real acceleration happens. Let’s build that future together. - **MikeIsaac** (2025-10-27): https://x.com/MikeIsaac/status/1982828494406263071 - Bruna Emerson Isaac 8/03/14 — 10/25/25 love you forever sweet girl https://t.co/PDJDPHuELz - **hichaelmart** (2025-10-27): https://x.com/hichaelmart/status/1982781872309727304 - Made Next.js fast. Er. • Bun: 1.15x faster • Node.js: 1.18x faster • CF Workers: 1.24x faster • Deno: 1.38x faster 🚀 https://t.co/sOVtH131wL - **hichaelmart** (2025-10-27): https://x.com/hichaelmart/status/1982780018095988786 - I've made a PR to optimize some of that work. It currently creates 1745 ReadableStreams per request unnecessarily 😬 This fix eliminates that and you get a 10-38% perf boost (Deno users get the most!) https://t.co/bdtH78O3dG - **hichaelmart** (2025-10-27): https://x.com/hichaelmart/status/1982780007358337310 - What makes Next.js slow? In cases like @theo's benchmark, it's because it's creating multiple React streams, parsing and rendering to JSON and HTML, reading the whole body along the way and doubling the payload. Next.js = 220ms, plain React streams the same thing in 27ms https://t.co/w7HNR0kt8E - **danshipper** (2025-10-27): https://x.com/danshipper/status/1982770929899114625 - I spend about 25% of my time in SF and I can assure you this is not true You can be just as on the edge living somewhere else—and have just as many inane 4-7 months behind conversations with AI people in SF https://t.co/GlJsKJyvjI - **doodlestein** (2025-10-27): https://x.com/doodlestein/status/1982631024774316220 - I finally got around to making a tool I've wanted for a long time: you can basically think of it as being "like Gmail for coding agents." If you've ever tried to use a bunch of instances of Claude Code or Codex at once across the same project, you've probably noticed how annoying it can be when they freak out about the other agent changing the files they're working on. Then they start doing annoying things, like restoring files from git, in the process wiping out another agent's work without a backup. Or if you've tried to have agents coordinate on two separate repos, like a Python backend and a Nextjs frontend for the same project, you may have found yourself acting as the go-between and liaison between two or three different agents, passing messages between them or having them communicate by means of markdown files or some other workaround. I always knew there had to be a better way. But it's hard to get the big providers to offer something like that in a way that's universal, because Anthropic doesn't want to integrate with OpenAI's competitive coding tool, and neither wants to deal with Cursor or Gemini-CLI. So a few days ago, I started working on it, and it's now ready to share with the world. Introducing the 100% open-source MCP Agent Mail tool. This can be set up very quickly and easily on your machine and automatically detects all the most common coding agents and configures everything for you. I also include a ready-made blurb (see the README file in the repo, link in the next tweet) that you can add to your existing AGENTS dot md or CLAUDE dot md file to help the agents better leverage the system straight out of the gate. It's almost comical how quickly the agents take to this system like a fish to water. They seem to relish in it, sending very detailed messages to each other just like humans do, and start coordinating in a natural, powerful way. They even give each other good ideas and pushback on bad ideas. They can also reserve access to certain files to avoid the "too many cooks" problems associated with having too many agents all working on the same project at the same time, all without dealing with git worktrees and "merge hell." This also introduces a natural and powerful way to do something I've also long wanted, which is to automatically have multiple different frontier models working together in a collaborative, complementary way without me needing to be in the middle coordinating everything like a parent setting up playdates for their kids. And for the human in the loop, I made a really slick web frontend that you can view and see all the messages your agents are sending each other in a nice, Gmail-like interface, so you can monitor the process. You can even send a special message to some or all your agents as the "Human Overseer" to give them a directive (of course, you can also just type that in manually into each coding agent, too.) I made this for myself and know that I'm going to be getting a ton of usage out of it going forward. It really lets you unleash a massive number of agents using a bunch of different tools/models, and they just naturally coordinate and work with each other without stepping on each other's toes. It lets you as the human overseer relax a bit more as you no longer have to be the one responsible for coordinating things, and also because the agents watch each other and push back when they see mistakes and errors happening. Obviously, the greater the variety of models and agent tools you use, the more valuable that emergent peer review process will be. Anyway, give it a try and let me know what you think. I'm sure there are a bunch of bugs that I'll have to iron out over the next couple days, but I've already been productively using it today to work on another project and it is pretty amazingly functional already! - **__syumai** (2025-10-26): https://x.com/__syumai/status/1982444815594955170 - Cloudflare Containersの登場で需要が無くなったかと思ったけど地味に伸び続けていて、なんか知らん間に年内1000 Starが見えてきたライブラリ情報です syumai/workers: Go package to run an HTTP server on Cloudflare Workers. https://t.co/IizjmarOrn - **thomas_ankcorn** (2025-10-25): https://x.com/thomas_ankcorn/status/1982207080422330478 - Yall are cute 🧡 https://t.co/9ZAzbozJQB https://t.co/bsfeRAXqot - **thdxr** (2025-10-25): https://x.com/thdxr/status/1982199365906108753 - i've seen so many case where there's a self serve automated system that a business eventually just hires someone to help customers use seen this in parking garages lately, there's the self pay ticket thing on exit but there's now a guy there who swipes your card for you - **southpolesteve** (2025-10-25): https://x.com/southpolesteve/status/1982178807655092361 - https://t.co/T7pxgpDO12 https://t.co/uIzNTMbuWz - **thdxr** (2025-10-25): https://x.com/thdxr/status/1982129382106816535 - because we built opencode as a client/server model from day one, 100% of it's functionality is available as an API this includes non-LLM functionality like searching for files this is going to pay off so well - **brian_scanlan** (2025-10-25): https://x.com/brian_scanlan/status/1982059875040739375 - Excellent article from @collision - optimistic and relevant commentary on Ireland's most important problem. https://t.co/ASsY8M7D6o - **jeresig** (2025-10-25): https://x.com/jeresig/status/1982046991132889191 - I've been using @tan_stack Start for a new project and it's super good. The server functions completely replace the need for TRPC/GraphQL/REST, the middleware is composable and fully typed, and having TSRouter's nice typing and stateful search params is icing on the cake. A+! - **mjackson** (2025-10-25): https://x.com/mjackson/status/1981921447158403165 - Devs who come to JavaScript from other languages are confused because everything is so centered around bundlers these days. It doesn’t need to be this way. We can make JavaScript much more friendly and approachable for new developers. https://t.co/fR6C4fXUcS - **dillon_mulroy** (2025-10-24): https://x.com/dillon_mulroy/status/1981858783610691696 - today was my last day at vercel. i got to help build the fastest domain search on the web, designed our new workflows db so it can scale to the moon, and most importantly worked with some of the best people in the world. taking a month off, but excited for what's next 😁 - **m0hilll** (2025-10-24): https://x.com/m0hilll/status/1981833780441887041 - @irvinebroque @CloudflareDev the upload fails but the spinner keeps loading. might wanna fix it. https://t.co/WqOqoW3nl4 - **elithrar** (2025-10-24): https://x.com/elithrar/status/1981817972768243881 - rarely a week goes by that I don't get to reference Hyrum's Law (@hyrumwright): > With a sufficient number of users of an API, it does not matter what you promise in the contract: all observable behaviors of your system will be depended on by somebody. - **mitsuhiko** (2025-10-24): https://x.com/mitsuhiko/status/1981798129466065401 - It’s interesting how firemen and soldiers are recognized by society universally for their services, but someone on call for a cloud service, gets no credit or visibility. Is it because they are paid better or is it just a thing of recency? - **tannerlinsley** (2025-10-24): https://x.com/tannerlinsley/status/1981764071897936369 - https://t.co/dYSm9mU6xE - **BraydenWilmoth** (2025-10-24): https://x.com/BraydenWilmoth/status/1981754888364441727 - Cloudflare dashboard got ~28% faster this week for LCP, with even more significant gains for users in high-latency regions. Also cut average time of loading screen visibility down by 1-2 seconds. - **cramforce** (2025-10-24): https://x.com/cramforce/status/1981741565157773377 - Huge miss in TS/JS that decorators don't work on functions. Is that in the works at all? - **simonw** (2025-10-24): https://x.com/simonw/status/1981724573692477882 - Geoffrey drops a new analogy for working with AI that I really like; you're the surgeon, the AI tools are your team of surgical assistants https://t.co/QX7MzCpwCi - **geoffreylitt** (2025-10-24): https://x.com/geoffreylitt/status/1981720163968749736 - A lot of people say AI will make us all "managers" or "editors"...but I think this is a dangerously incomplete view! Personally, I'm trying to code like a surgeon. A surgeon isn't a manager, they do the actual work! But their skills and time are highly leveraged with a support team that handles prep, secondary tasks, admin. The surgeon focuses on the important stuff they are uniquely good at. My current goal with AI coding tools is to spend ~100% of my time doing stuff that matters. (As a UI prototyper, that mostly means tinkering with design concepts.) It turns out there are a LOT of secondary tasks which AI agents are now good enough to help out with. Some things I'm finding useful to hand off these days: - Before attempting a big task, write a guide to relevant areas of the codebase - Spike out an attempt at a big change. Often I won't use the result but I'll review it as a sketch of where to go - Fix typescript errors or bugs which have a clear specification - Write documentation about what I'm building I often find it useful to run these secondary tasks async in the background -- while I'm eating lunch, or even literally overnight! When I sit down for a work session, I want to feel like a surgeon walking into a prepped operating room. Everything is ready for me to do what I'm good at. --- Notably, there is a HUGE difference between how I use AI for secondary vs primary tasks. For the core design prototyping work, I still do a lot of coding by hand, and when I do use AI, I'm more careful and in the details. I need fast feedback loops and good visibility. (eg, I like Cursor tab-complete here) Whereas for secondary tasks, I'm much much looser with it, happy to let an agent churn for hours in the background. The ability to get the job done eventually is the most important thing; speed and visibility matter less. Claude Code has been my go-to for long unsupervised sessions but Codex CLI is becoming a strong contender there too, possibly my new favorite. These are *very* different work patterns! Reminds me of @karpathy's "autonomy slider" concept. It's dangerous to conflate different parts of the autonomy spectrum -- the tools and mindset that are needed vary quite a lot. --- The "software surgeon" concept is a very old idea -- Fred Brooks attributes it to Harlan Mills in his 1975 classic "The Mythical Man-Month". He talks about a "chief programmer" who is supported by various staff including a "copilot" and various administrators. OK, so there is a super obvious angle here, that "AI has now made this approach economically viable where it wasn't before", yes yes... But I am also noticing a more subtle thing at play, something to do with status hierarchies. A lot of the "secondary" tasks are "grunt work", not the most intellectually fulfilling or creative part of the work. I have a strong preference for teams where everyone shares the grunt work; I hate the idea of giving all the grunt work to some lower-status members of the team. Yes, junior members will often have more grunt work, but they should also be given many interesting tasks to help them grow. With AI this concern completely disappears! Now I can happily delegate pure grunt work. And the 24/7 availability is a big deal. I would never call a human intern at 11pm and tell them to have a research report on some code ready by 7am... but here I am, commanding my agent to do just that! --- Finally I'll mention a couple thoughts on how this approach to work intersects with my employer, @NotionHQ First, as an employee, I find it incredibly valuable right now to work at a place that is bullish on AI coding tools. Having support for heavy use of AI coding tools, and a codebase that's well setup for it, is enabling serious productivity gains for me -- *especially* as a newcomer to a big codebase. Secondly, as a product -- in a sense I would say we are trying to bring this way of working to a broader group of knowledge workers beyond programmers. When I think about how that will play out, I like the mental model of enabling everyone to "work like a surgeon". The goal isn't to delegate your core work, it's to identify and delegate the secondary grunt work tasks, so you can focus on the main thing that matters. - **boristane** (2025-10-24): https://x.com/boristane/status/1981716869393359266 - woke up today and decided to do wide design review of the workers observability dashboard great things coming - **dillon_mulroy** (2025-10-23): https://x.com/dillon_mulroy/status/1981464293443850564 - i wonder if as agents become more and more common place we'll also see the rise of HATEOAS style link relations in APIs outside of MCP servers for natural discovery of data and actions https://t.co/LAjJmWIrOu - **tannerlinsley** (2025-10-23): https://x.com/tannerlinsley/status/1981388215161634821 - I'll continue to die on the hill of "magic directives are bad API interfaces". They're not type-safe, they're not extensible, no native runtime control, they're just strings. ... I can't believe I'm saying this, but I would rather debate @ directives. Pls rewind history. - **nandafyi** (2025-10-23): https://x.com/nandafyi/status/1981375182435745918 - some design fixes on the workers metrics tab - scrollable overflow to the right and styled selects (after/before) https://t.co/My55Bko0ES - **aaronholla_** (2025-10-22): https://x.com/aaronholla_/status/1981064738155995309 - What if you could render react-native apps fully on the server without any bundler tricks? cc/ @zebassembly @csparks1919 https://t.co/xkoYY47T3d https://t.co/L59wg5uzsd - **akaphill** (2025-10-22): https://x.com/akaphill/status/1981029864657314194 - workflow builder playground - create workflows from a prompt and instantly deploy to cloudflare with all bindings and resources built this over the weekend for fun. should we ship? https://t.co/gXSwJiWCYi - **fatih** (2025-10-22): https://x.com/fatih/status/1980914664235040891 - I was a huge advocate for having e2e/integration tests tied to Pull Requests, but I no longer believe that's a great idea. e2e tests by nature are very finicky. They are long and have complex interactions. On top of it, they almost never run with the same stack that Prod is running (you can argue that you have prod similar environments, but nope, it'll be never the same). What happens with time is, because of the underlying assumptions and long tests, the tests start getting flakey. A 3-4 min CI run could end up being 30 min in a few years, and you might think that's the norm. It's not. So what is a better solution? Separate CI and CD. Merge changes only with unit tests, or better tests that passes under 2 minutes. Have a `async` e2e CI pipeline that runs on your main/master branch. Deployment to production is only allowed when the e2e passes. The benefits are immense: * It increases velocity. * Happy engineers * Increases product ship rate * Still have e2e but it no longer blocks you. You might think it's postponing and is the same. It's not. Imagine you're building a car. With the previous approach, everytime you make a change in your car, you need to drive it several miles. Even for small changes. With the later, engineers can focus on shipping the car, add feature, bug fixes etc..., without blocking people working on the car. At @PlanetScale we have multiple repositories and products, there is no one rule. But we have a dedicated 7/24 hour e2e lanes that runs constantly, always. If the e2e lanes are unhappy, we're looking into it. One of the e2e lanes always creates databases, and we have a randomizer that constantly changes certain database parameters. It's basically a self-feeding control loop. That allows us to check if rollouts are working as intended, and if the he database is always up and running. It allows us to catch things that are not possible because with e2e that are created from scratch on each Pull Request, because in the e2e pipeline, the database is never deleted, and it always carries state from the previous rollout. We don't purge these, unless we do major e2e test changes. Of course we also have e2e lanes that do different things, but just wanted to show howe use it. - **kathyyliao** (2025-10-22): https://x.com/kathyyliao/status/1980826304984543565 - @irvinebroque @ericclemmons 👋 dm-ed you @ericclemmons - **dexhorthy** (2025-10-21): https://x.com/dexhorthy/status/1980670791500095855 - Today, I'm humbled and honored to to officially extend to @0xblacklight a VERY WARM WELCOME as CTO and CoFounder at HumanLayer. I couldn't be more stoked to be on this journey w/ Kyle and if you thought @HumanLayer was crushing it before...you ain't seen nothing yet https://t.co/WeWAVwcqat - **nabukhotnyi** (2025-10-21): https://x.com/nabukhotnyi/status/1980651826526711853 - Drizzle Studio Chrome Extension now supports @Cloudflare Durable Objects. https://t.co/HvGvXF9art - **nandafyi** (2025-10-21): https://x.com/nandafyi/status/1980650366913163460 - New post 🎉 Going back to my roots on writing about the inner workings of things, a breakdown of key-value databases and how you might make one from scratch: https://t.co/Dpy0TGQpZO https://t.co/Oz2YYCe2I1 - **GergelyOrosz** (2025-10-21): https://x.com/GergelyOrosz/status/1980645159475007779 - So ridiculous. AWS didn't "break the entire internet." Companies that decided to build non-resilient systems depending on one single cloud region (aws-us-east-1) broke themselves. Noticed how X, Google, Meta, Shopify etc were all fine? "Breaking up" would solve nothing... https://t.co/FP7WOO0tPt - **boristane** (2025-10-21): https://x.com/boristane/status/1980444978406097048 - time for the "crm for your twitter dms" crowd to shine https://t.co/OWvMUCozUW - **dok2001** (2025-10-20): https://x.com/dok2001/status/1980413815234916739 - @MrAhmadAwais @CloudflareDev @irvinebroque good idea - **zlxndr** (2025-10-20): https://x.com/zlxndr/status/1980381645417910758 - I may be anchored to what i’ve seen before on other platforms but I find it most easy to navigate the ui, if: - there is a single service - which has n environments I create (lets say prod & preview) - to me the fact that environments are actually 2 workers under the hood is exactly the impl detail I don’t really care about (only that it works) I usually check if both environments have the right variables set and it would be useful to be able to quickly cycle through them instead of going back and forth - **EastlondonDev** (2025-10-20): https://x.com/EastlondonDev/status/1980356125817204879 - I kinda want to run my own email, docs, GitHub and DockerHub on cloudflare. Seems like it wouldn’t be too bad at this point - **zeeg** (2025-10-20): https://x.com/zeeg/status/1980317592100413668 - Was chatting with @threepointone about MCP challenges (mostly complexity related)... We're going to fully drop SSE on Sentry's MCP server. Now is your last chance to lodge a complaint before I gut this thing. - **threepointone** (2025-10-20): https://x.com/threepointone/status/1980312388500115753 - @wesbos capnweb? - **gching** (2025-10-20): https://x.com/gching/status/1980297959188148366 - shipping @OpenAIDevs ChatGPT apps on @CloudflareDev is way too easy 🚀 built Wordle in ChatGPT and its 🔥 the ChatGPT + Cloudflare combo is the future wanna ship your own? i built a template here: https://t.co/94XkuU8ge0 next? making a template for mcp-ui @liadyosef @idosal1 https://t.co/DQTFjcOsRs - **sidi_jeddou_dev** (2025-10-20): https://x.com/sidi_jeddou_dev/status/1980291269101298148 - Introducing next-to-tanstack: a CLI to migrate your Nextjs projects to @tan_stack start I came across TanStack for the first time through @ImSh4yy a few months ago, and really liked the ecosystem and the idea of how you can deploy it easily on: -@Cloudflare -@Netlify …etc More and more people are moving away from Nextjs to TanStack, so I think this framework will be huge the next few years. This OSS project is still in progress. The goal is to make migrating from Nextjs as easy and time-saving as possible. The repo is in the comments, all PRs are welcome 👇 - **ritakozlov** (2025-10-20): https://x.com/ritakozlov/status/1980278941332951425 - at this year's @cloudflare investor day i talked about why cloudflare is well-positioned to become the AI cloud for the next gen of applications we've built out all the primitives you need to ship AI applications, and by running on region: earth give you: 1. scalability & cost 2. performance 3. dev exp a few updates since then... containers open beta & sandboxes being able to run untrusted code is going to be an essential part of the agentic future for many reasons. first of all, code gen is the first agentic use case that's achieved PMF. for this you need to be able to check out code and run it in secured environments. additionally, personalized software is on the rise, which means needing to scale up horizontally to support many smaller applications in addition to vertically. code-mode speaking of agents and running untrusted code. we've found that many LLMs still struggle with tool-calling, but are really good at writing code. enabling LLMs to run their executive functions through code unlocks a lot of potential, which is why we're working on enabling workers to support rapidly importing and running code on demand, without uploading it pre-emptively. workflows GA many AI powered workloads today follow more of a workflow than an agentic execution pattern. meaning, running a set of steps in a durable-execution environment. cloudflare workflows makes this easy and possible data platform during birthday week, cloudflare launched our data platform which builds on R2 by making streaming ingestion of data into R2 AND querying over data in R2 easy leveling up the platform on top of all the new products, we've made significant investments in: - workers ai (our inference platform), which now runs more models, more efficiently than ever. - ai gateway: which allows access to all models (including openai, anhtropic, etc) with unified billing, advanced routing funtionality, and better observability putting it all together with agents sdk we launched agents sdk to make it easy to put together all the pieces to build an agent, including: - memory: state management and memory is made easy to with agents sdk by building on durable objects - multi-modality: chat agents are just step 1 — though agents sdk comes with built-in websocket management. next phases for agents are going to include voice, email. agents sdk makes this easy. - MCP: all the top remote mcp servers you see are built on agents sdk... all of this is built on region: earth by default, which means you get the best of: performance, scalability, security and connectivity to all the piecs you need can't stop, won't stop — importantly, we're still just getting started - **RichOBray** (2025-10-20): https://x.com/RichOBray/status/1980276606334861693 - Full video https://t.co/WNAEs8bOsM - **RichOBray** (2025-10-19): https://x.com/RichOBray/status/1980000065105101123 - Been messing around with Cap'n Web by @KentonVarda and man, this bi-directional calling stuff is awesome. Planning to release a video about it next week. https://t.co/QSYNWMIdDo - **yusukebe** (2025-10-18): https://x.com/yusukebe/status/1979685460957102179 - When I started writing Hono 4 years ago, I was alone. Now, Hono has 26K starts, 9M/mo downloads, 687 contributors. Yesterday. We did Hono Conference. 184 came in person. 184? Who was expecting this? 184 people were there for my software, which I wrote alone. Tomorrow never knows https://t.co/jBXI6rKZp7 - **threepointone** (2025-10-18): https://x.com/threepointone/status/1979533505362002264 - (not a joke) someone needs to make a natural language linter (an "editor", if you will) that helps make the writing clear and ambiguous, tweaked for specifying actions etc (this part is a joke) we can call it a programming language https://t.co/KB1YQ38swI - **csparks1919** (2025-10-18): https://x.com/csparks1919/status/1979346587869679853 - what do you mean I can't useState in a server component? https://t.co/5pNOHIIVcg https://t.co/USJ4zdTMS2 - **southpolesteve** (2025-10-17): https://x.com/southpolesteve/status/1979285274137104470 - Saw a demo of this today. Absolutely blew my mind. Think RSC but without a bundler. Render on the client or the server. Event handlers on the client or the server. All at runtime via Cap'n Web. Oh and it can work with any framework. Not just React https://t.co/gE2p5xHPJH - **martin_casado** (2025-10-17): https://x.com/martin_casado/status/1979263667507204447 - It may be the case that the primary unlock of this AI wave is not generalized superintelligence, but the ability to pour lots of resources into a given problem domain effectively. In the past this was very difficult because most solutions reduced to software engineering which is notoriously difficult to accelerate with additional capital. Overfunding has historically been a great way to ruin a great project or company. However, these models really are universal function approximators. And as an industry we've refined the ability to apply additional capital to the training (pre and post) to get better and better solutions in a very broad set of domains (language, science, code, math, robotics, etc.). Whether or not a single model with general super intelligences arises from this, the impact could be very similar -- a systemized way to directly apply large amounts of capital to problem domains. And in that world the primary limiter to solving any suitable problem is the economic value of the problem being solved. Or the collective willingness to solve it regardless. - **dwarkesh_sp** (2025-10-17): https://x.com/dwarkesh_sp/status/1979234976777539987 - The @karpathy interview 0:00:00 – AGI is still a decade away 0:30:33 – LLM cognitive deficits 0:40:53 – RL is terrible 0:50:26 – How do humans learn? 1:07:13 – AGI will blend into 2% GDP growth 1:18:24 – ASI 1:33:38 – Evolution of intelligence & culture 1:43:43 - Why self driving took so long 1:57:08 - Future of education Look up Dwarkesh Podcast on YouTube, Apple Podcasts, Spotify, etc. Enjoy! - **csparks1919** (2025-10-17): https://x.com/csparks1919/status/1979215622027514284 - interactive server components have been acheived internally https://t.co/f8Ie0kmbiv - **cramforce** (2025-10-17): https://x.com/cramforce/status/1979203221941109234 - We're celebrating the life of @mikeal, creator of request, forker and reuniter of Node.js, renaissance and family man on November 12. We received a very generous donation and are now able to make this event completely free. Looking forward to see all of you! https://t.co/eBF6qK6wfC - **fatih** (2025-10-17): https://x.com/fatih/status/1979153779967569988 - I'm going to die on this hill, but this is exactly my thinking. NO ONE SHOULD CARE ABOUT THE MODEL. It's not my job to understand the differences. Do you know which version Google is using for their search, or which version your favorite App is running? No. Why should I care which model my Agent is running? Currently both @AmpCode and @cursor_ai got it correct. Amp doesn't advertise and just calls it "smart", and Cursor comes with "auto", which is basically telling you "just code, don't think about the model, we'll give you the best user experience". So, the agent should be the doing what it is at the best, meaning it should be able to understand my prompt properly, and execute it. Does it run Sonnet 4.5? So what? Who cares. All I care is the output. If I don't like the output of it, I'm free to use other agents. - **grinich** (2025-10-17): https://x.com/grinich/status/1979034812317888916 - my favorite demo from AI Tinkerers tonight: @andrewtorkbaker hacking Claude for Chrome to control 90s drawing app Kid Pix 🎨 also finally got to meet @GergelyOrosz who is visiting SF! https://t.co/0gnKKJEsS6 - **hkfoster** (2025-10-16): https://x.com/hkfoster/status/1978969464520552680 - Working on it 💅 https://t.co/IJpzGUWiS2 - **Jilles** (2025-10-16): https://x.com/Jilles/status/1978955632305733888 - OK Cloudflare Sandboxes are *really cool*. In a few hours I was able to spin up a TanStack Start application that sends Go code to a Worker using the Sandbox SDK and runs some tests. WILD TIMES https://t.co/BJNK5C2NXY - **simonw** (2025-10-16): https://x.com/simonw/status/1978943056889156064 - @kristoph That's why I think it's so cool - adding new capabilities to your LLM system by dropping in a single Markdown file is a delightfully elegant solution - **wishee0** (2025-10-16): https://x.com/wishee0/status/1978942919475126594 - HELL YEAH! I figured out realtime updates for workflows, coolest thing i built today! inject instance start/stop logs for nodes -> tail -> stream and update the ui cf employees were insanely helpful here for ideating 🧡!!! https://t.co/zDK6Qn1Bas https://t.co/YmG931blgq - **dok2001** (2025-10-16): https://x.com/dok2001/status/1978923289390256273 - We won’t stop until the dashboard is rebuilt! https://t.co/VYlo4pEoI3 - **wesbos** (2025-10-16): https://x.com/wesbos/status/1978912288598794268 - cloudflare can't keep getting away with this these designs are SO GOOD https://t.co/rs5ASnaxGu - **latentsaint** (2025-10-16): https://x.com/latentsaint/status/1978897727741133277 - I am officially GEPA and @DSPyOSS pilled. Im building a doc -> json for a bespoke document class, and this raised me from 73% baseline to 98% on a three stage pipeline using gpt-4.1-mini - **samlambert** (2025-10-16): https://x.com/samlambert/status/1978874553557488039 - I don't think people realize that companies Block (Cash App), Intercom, Cursor, Whop, MyFitnessPal, Kick run their largest databases ON OUR CLOUD. This is a massive responsibility that we take very seriously. It's also incredibly different from selling on prem software. - **astuyve** (2025-10-16): https://x.com/astuyve/status/1978872281012011433 - @irvinebroque @mibullock you would not believe the horrors I've seen in tracestate - **MikeNomitch** (2025-10-16): https://x.com/MikeNomitch/status/1978868988437500236 - Very happy to have a SQL Editor on Durable Objects now 🤩 (No more writing custom methods for o11y on my DOs!) https://t.co/37GY9p9Fq7 https://t.co/Vpz8L46Pqc - **KentonVarda** (2025-10-16): https://x.com/KentonVarda/status/1978848606644752417 - You can now remote query your Durable-Object-embedded SQLite databases from inside the Cloudflare dashboard UI. https://t.co/06xnjs64zd - **BraydenWilmoth** (2025-10-16): https://x.com/BraydenWilmoth/status/1978836640211546476 - Data Studio for Durable Objects is now... live! I might have been a bit off on my "next week" estimate but we made it happen. Huge shoutout to everyone involved – was a cross-team effort. Big thanks to @KentonVarda for making it all possible. https://t.co/Yh9gBnccmU - **dok2001** (2025-10-16): https://x.com/dok2001/status/1978815948443300306 - The level of actionable observability in @PlanetScale is unreal. https://t.co/QlyXKIVFnd - **yusukebe** (2025-10-16): https://x.com/yusukebe/status/1978788894629105671 - I met a Hono friend @MathurAditya7 https://t.co/i1XyTLItn4 - **zero__ms** (2025-10-16): https://x.com/zero__ms/status/1978770591260787109 - 🎃 🎃 Zero 0.24 is out! Featuring: * Faster `whereExists()` via "join flipping" * Simpler cookie-based auth * Preview support * Query analyzer built into inspector * Faster SQLite * New full-stack @Cloudflare demo ... and a *ton* more. Details at: https://t.co/lyHfokchVY - **SamNewby_** (2025-10-16): https://x.com/SamNewby_/status/1978689389489168814 - holy shit I need to try Cloudflare sandbox - **michellechen** (2025-10-15): https://x.com/michellechen/status/1978512320390115432 - new workers ai partnerships and demos at cf connect keynote 🧡 1) @IBMDeveloper granite model, now on workers ai 2) firewall for ai demo, powered by workers ai 3) voice agents demo, powered by @DeepgramAI models on workers ai https://t.co/qWkO9nS8yO - **dinasaur_404** (2025-10-15): https://x.com/dinasaur_404/status/1978509875534598155 - they did it again! new @Cloudflare site just dropped 👀 if you're building an AI vibe coding platform or AI agents that needs to execute code, sandboxes enable you to run the untrusted code in a secure, isolated environment 👇 https://t.co/7BNNpS8Z9e - **nandafyi** (2025-10-15): https://x.com/nandafyi/status/1978505542269059208 - drafted this design a while back, super excited to see it come to life! kudos @whoiskatrin and team :) https://t.co/iAs7RDMiUE - **samlambert** (2025-10-15): https://x.com/samlambert/status/1978501354344636435 - If you are at #cloudflareconnect i am giving a talk on Thursday morning where I will be sharding an application live. - **whoiskatrin** (2025-10-15): https://x.com/whoiskatrin/status/1978500458168664173 - say hello to Sandboxes ⚡ run any command or process, expose URLs, make file operations, stream real-time logs and directly execute python & javascript! spin up secure & scalable sandboxes with a simple SDK https://t.co/JMxLzdtI0w https://t.co/M3BR0x2h6e - **grinich** (2025-10-15): https://x.com/grinich/status/1978465132289101917 - Enterprise Ready Conf is next week! 🔥 We have some killer speakers from @Rippling, @Webflow, @Render, @getpostman and more Plus some big new launches from the @WorkOS team 👀 See you there! https://t.co/RKb7ywajla - **leinwand** (2025-10-15): https://x.com/leinwand/status/1978432363265687598 - Great to see @Cloudflare move fast to make the Internet faster for everyone. @webflow hosted sites and @webflow cloud got even faster with these changes. Love the optimization on JSON.parse() for everyone using V8. Thanks @KentonVarda @dok2001 and team! 🙌🙏 https://t.co/kQIa42G1Lh - **tmikov** (2025-10-15): https://x.com/tmikov/status/1978340173084496117 - This is pretty awesome. A rare example of how things are supposed to work. https://t.co/AUFHTCudVE - **liadyosef** (2025-10-15): https://x.com/liadyosef/status/1978260902894838236 - An MCP-UI app. Rendering natively inside ChatGPT. Built with @remix_run by @kentcdodds. And with an homage to Tron! It just can't get any awesomer. What a future we live in! 🚀 (cc @idosal1) https://t.co/fIIrEQSTlX - **0x15f** (2025-10-14): https://x.com/0x15f/status/1978230657928106227 - This is an absolutely incredible read. I highly recommend anyone and everyone read it, review the benchmarks, and review the PRs. Not only did Cloudflare get faster but the web, nodejs, and entire js ecosystem was improved https://t.co/lEivpNdtir - **kentcdodds** (2025-10-14): https://x.com/kentcdodds/status/1978219213904044051 - This is my demo for #CloudflareConnect. I made it so when the user's calculation results in 1982 (the year the original Tron was released), ChatGPT receives a prompt instructing it to behave like the Master Control Program (MCP) from the movie 😆 Built with Remix 3. https://t.co/TUJno5p5Vn - **MikeNomitch** (2025-10-14): https://x.com/MikeNomitch/status/1978217915779252465 - Was an absolute pleasure to watch Cloudflare engineers go heads down last week to make Workers faster. Huge speed improvements for Workers (and some real performance improvements upstreamed to Node and V8 at the same time) https://t.co/wOZgwHaT5u - **janwilmake** (2025-10-14): https://x.com/janwilmake/status/1978207429985747044 - @dok2001 @theo thanks, I can finally stop using vercel completely! https://t.co/aZHT7csERr - **dok2001** (2025-10-14): https://x.com/dok2001/status/1978204911339995640 - Team had a lot of fun making workers faster. Big thank you to @theo for pushing us to be better! https://t.co/F0rA7MuEwt - **zebassembly** (2025-10-14): https://x.com/zebassembly/status/1978203727564792175 - I spent much of last week helping investigate performance issues and running benchmarks from an EC2 VM and it was an incredibly fun experience. The team behind the Workers runtime is world class. One of my favorite parts about this is when we noticed an issue impacting Vercel and making us look better than we should have been. Despite the fact it'd make it harder for us to win the benchmarks, an engineer said "a bug is a bug and we fix bugs" and now we've submitted a PR to Node to give customers on Vercel (and anyone using node in general) some nice perf gains. - **janwilmake** (2025-10-14): https://x.com/janwilmake/status/1978203597902143855 - holy shit! I built uithub on Vercel more than a year ago, and built a v2 on workers early this year, but never finished it because it was slower. I always assumed it was because of memory constraints (128MB), didn't even think about it being the CPU. Just tested this and now it's more than 2x faster 🤯 MIND BLOWN In below example: - uuithub uses Cloudflare with 128MB RAM = 9.45s - uithub uses Vercel with 2GB RAM = 21.11s Time to finis that migration? - **boristane** (2025-10-14): https://x.com/boristane/status/1978201725791056277 - workers got faster this means all other products that leverage workers also got faster 🤯 https://t.co/AUQZidf3CB - **theo** (2025-10-14): https://x.com/theo/status/1978200417516355983 - Cat's out of the bag. Cloudflare is ~3x faster for many workloads due to a series of fixes they've shipped in the last 5 days https://t.co/SnLpLzMUnX https://t.co/fJhss9TZGu - **ritakozlov** (2025-10-14): https://x.com/ritakozlov/status/1978198338265653352 - we didn’t like the benchmark results so we made @cloudflare workers faster & set the record straight⚡️ here’s how 👇 https://t.co/EQvlwJaJm2 https://t.co/5up9OddbgO - **whoiskatrin** (2025-10-14): https://x.com/whoiskatrin/status/1978197780230238250 - there is magic in making JavaScript faster https://t.co/MzMmWWTokq - **yagiznizipli** (2025-10-14): https://x.com/yagiznizipli/status/1978197671132242151 - My kind of blog post https://t.co/SwTYLE7Xfd - **KentonVarda** (2025-10-14): https://x.com/KentonVarda/status/1978197423085306100 - Those benchmarks from @theo revealed some interesting issues in Workers. Happy to say it's all fixed now, save for some lingering next.js-specific stuff we're continuing to work on. Thanks for the reproducible test cases, @theo. (Link to blog post in reply.) https://t.co/5gIIdRQSvc - **threepointone** (2025-10-14): https://x.com/threepointone/status/1978196474895950233 - Put your notifs off, take 10 minutes and read this. As a JavaScript nerd this was amazing to see come together. We kinda made the internet faster this week. https://t.co/2IE4jNf5yw - **Cloudflare** (2025-10-14): https://x.com/Cloudflare/status/1978190852078887182 - Cloudflare investigated CPU performance benchmark results for Workers, uncovering and fixing issues, making Cloudflare Workers faster for all customers. https://t.co/uxUmNKgTea - **thdxr** (2025-10-14): https://x.com/thdxr/status/1978161249008353370 - devrel seems to die when money dries up devrel seems to come back when money is cheap almost like you can conclude something from that - **wishee0** (2025-10-14): https://x.com/wishee0/status/1978160610849550406 - building a visual workflow builder where you drag and drop nodes for @Cloudflare workflows and it instantly generates real valid worfklows. it already supports kv, d1, http calls, conditionals, transforms, etc. finally got the first end-to-end run to work. right now, i have to manually deploy the generated code and bindings next: auto-deploy + workers tail for displaying live execution. later, i want agents that suggest nodes and generate workflows on the fly, maybe even spin up containers to validate the code automatically. wild that this didn’t already exist, so i'm making it. - **kentcdodds** (2025-10-14): https://x.com/kentcdodds/status/1978158204736430234 - Had a use case today where I wanted to update a value without triggering a rerender and I love that all I had to do was not call this.update() - **kentcdodds** (2025-10-14): https://x.com/kentcdodds/status/1978152141119189321 - I've got such a fun demo for my talk at #CloudflareConnect today. Speaking in 45 minutes. It's gonna be a hoot! - **williamallen** (2025-10-14): https://x.com/williamallen/status/1978094745323774344 - Two challenges agents & merchants have today: 1) How do you authorize an agent to represent you and use your credit card? 2) How do you let a merchant know that the agent *really does* represent you and isn't a random bot? That's what @Visa, in collaboration with @Cloudflare, is helping to solve with the Trusted Agent Protocol. The Trusted Agent Protocol uses HTTP Message Signatures to allow networks like Cloudflare to verify traffic from agents that register with Visa on behalf of a cardholder. This allows merchants to distinguish legitimate, approved agents from random bots. You can start to feel the promise of agentic commerce, but it hasn't arrived yet. I think this takes us another step closer. - **burcs** (2025-10-13): https://x.com/burcs/status/1977837367496216679 - @_skris @Cloudflare yeah you're right we should have this, will add it to the list, thanks for calling it out - **samgoodwin89** (2025-10-13): https://x.com/samgoodwin89/status/1977822596684300572 - Remix 3 looks like a win to me. Implicit -> explicit is often the right move. Simplicity over magic. - **korinne_dev** (2025-10-13): https://x.com/korinne_dev/status/1977800556648489441 - Moved my home up to the Bay Area! Felt unfairly romantic to sail under the Golden Gate Bridge. https://t.co/UAQeDrmRWd - **_skris** (2025-10-13): https://x.com/_skris/status/1977735926718599296 - there should be a "set as default domain" option @Cloudflare and then redirect to preferred domain automatically https://t.co/BBVr8jDnfJ - **Steve_Yegge** (2025-10-13): https://x.com/Steve_Yegge/status/1977645937225822664 - I just released Beads, a drop-in cognitive upgrade for your coding agent of choice. https://t.co/d94hykIk7y In a nutshell, it is a magical 4-dimensional graph-based git-backed fairy-dusted issue-tracker database, designed to let coding agents track all your work and never get lost again. Beads replaces that disgusting pile of half-eaten markdown files in your plans/ directory. You know you have one. Install the Beads binary, tell your agent in https://t.co/q9aIg3VyQZ to stop using Markdown and run `bd quickstart`, and your agents will spontaneously get better at everything, particularly long-horizon planning and keeping track of newly discovered work. If you are looking for a long-ass but occasionally funny article about how I discovered it while burning a 350k line code base to the ground: https://t.co/DQ0ZZIJhRT Or you can, you know, just start using it. It's MIT-licensed, written in Go, and super lightweight and portable. Let me know how you like it. Enjoy! - **mjackson** (2025-10-13): https://x.com/mjackson/status/1977563966114259061 - Almost nobody is talking about the router we showed at Remix Jam but I think it’s the best router we’ve ever built. There is a lot to unpack here, and I’ll be writing docs this week to try and illustrate just how cool it is. https://t.co/6dT9d3t7Ek - **DanielLurie** (2025-10-12): https://x.com/DanielLurie/status/1977431115578364177 - There are tens of thousands of people in San Francisco this weekend for concerts, Fleet Week, and the Italian Heritage Parade. This is what a world-class city looks like, and it all starts with public safety. Crime is down 30% and tent encampments are at an all-time low. San Francisco is on the rise. - **matthewcp** (2025-10-12): https://x.com/matthewcp/status/1977400220813078978 - @oriSomething @ryanflorence @mjackson this.addEventListener('props', newProps => { name = https://t.co/oFZTKdguA5; this.update(); }) Or make a `props` import or something, shape doesn't matter to me. - **pcattori** (2025-10-12): https://x.com/pcattori/status/1977364285555700222 - One thing most people don't realize is that 3rd party libraries shouldn't need to add special support for Remix 3. E.g. we don't https://t.co/ekl7wQXg1b to have a "Remix" option like they do for React, Vue, etc. You should be able to use the JS version directly. No need for a wrapper library like `remotion`. I think this will be possible thanks to the native events-based model. So some libraries may work out-of-the-box with Remix 3, and if not they should just need to add events that _all_ frameworks — even vanilla JS — would benefit from. That's the vision for now, but we'll see how close to that we can get. Not sure what you mean by "DOM interactivity testing", but any DOM/browser testing library — Playwright, Vitest, etc. — should work with Remix 3. There's no "adapter layer" necessary for that. For UI libraries: yes, there are a couple essential UI libraries that React has that Remix 3 will need. That's one big reason why we are going to build a component library / design system ala shadcn. So yes, there's a gap here that needs to be filled, but I think that its not as big of a gap as you might think thanks to the events-based interop I mentioned at the beginning. - **juciano_barbosa** (2025-10-12): https://x.com/juciano_barbosa/status/1977359390907813916 - Remix 3 is truly just JavaScript https://t.co/2chYuvCx9y - **mjackson** (2025-10-12): https://x.com/mjackson/status/1977239524435738880 - Use the platform. - native events - abort signals - built-in css support - HTML over the wire - seamless support for web components That’s Remix UI. - **darkmembo** (2025-10-12): https://x.com/darkmembo/status/1977201025275904155 - @davidhoang 👀 - **MarcSelwan** (2025-10-12): https://x.com/MarcSelwan/status/1977187942314783209 - @irvinebroque Any app that doesn’t respect CMD + K for navigation is dead to me - **adamwathan** (2025-10-12): https://x.com/adamwathan/status/1977172480264720585 - New Cloudflare Workers site is beautiful! ✨ Awesome work by @tjcages. Built with Tailwind CSS v4 💅🏻 https://t.co/dRTZ4cwxnY - **soleio** (2025-10-12): https://x.com/soleio/status/1977170498959413321 - I have a way to solve declining birth rates at a global scale but no one will take it seriously https://t.co/Wb9j6aTNsJ - **burcs** (2025-10-11): https://x.com/burcs/status/1977156342398808377 - @realguse we're working on it, have a section in particular you'd like to see next? https://t.co/F42th1ZGDm - **ryanflorence** (2025-10-11): https://x.com/ryanflorence/status/1977155662397243792 - @thdxr and it's not really even different 😅 "React knows to update because I called a function but Remix only knows to update if I call a function" - **pcattori** (2025-10-11): https://x.com/pcattori/status/1977153071659581619 - I couldn't ask for a bigger compliment than "Remix 3 is so simple I already understand it well enough to critique it" For better or worse, @remix_run 3 is radically simple. I think better, but curious what you think. - **ryanflorence** (2025-10-11): https://x.com/ryanflorence/status/1977146888559182304 - Hey, I know you think this.update() is manual updates but setState for some reason isn't (?) there's something more interesting. Did you see that frame.reload() reconciled AN HTML RESPONSE against both the REAL DOM and the HYDRATED COMPONENTS?! - **zebassembly** (2025-10-11): https://x.com/zebassembly/status/1977137622645342508 - Saw remix 3 and I still haven't formed an opinion around the ideas, but I'm liking the minimal bundling approach. No vite, no webpack, just esbuild for JSX and minification. - **dok2001** (2025-10-11): https://x.com/dok2001/status/1977135390973976944 - . @Cloudflare is your Security Cloud Performance Cloud Network Cloud Developer Cloud Media Cloud Realtime Cloud Data Cloud AI Cloud Agent Cloud - **ryanflorence** (2025-10-11): https://x.com/ryanflorence/status/1977126665395548275 - @ryanleichty @evanyou bundlers are great tools but, remix 3 is a runtime framework, not a compiled framework, so ... https://t.co/sYxsHv1S4p - **ritakozlov** (2025-10-11): https://x.com/ritakozlov/status/1977094907686010940 - claudia cardinale + diane keaton 🤍 two such huge powerhouses of cinema in rapid succession. RIP 😭 - **Baconbrix** (2025-10-11): https://x.com/Baconbrix/status/1977069976000536582 - These guys just keep innovating and making the React space better! https://t.co/5LKESgvOcb - **svensauleau** (2025-10-11): https://x.com/svensauleau/status/1977069319071863095 - From cloudflare worker to wood worker. https://t.co/nk7WKKbpIl - **kane** (2025-10-11): https://x.com/kane/status/1977067561679553007 - The strange saga of Marc Benioff: > goes full-Woke in 2010s, donates millions to support Prop C > Prop C is an ill-designed gross receipts tax to subsidize progressive pro-drug policies > directly causes biz to leave SF and drug traffickers to move in > Marc leaves for private Hawaiian ranch > junkies and traffickers ruin the area around his office > Marc returns to the disgusting state of downtown > 2025 Dreamforce, needs to be nice > asks for the National Guard to fix the mess he had a direct hand in creating > gets yelled at on the internet - **kvlly** (2025-10-11): https://x.com/kvlly/status/1977067482214285404 - Legend has it gym memberships are particularly hard to cancel here because nobody knows how to exit https://t.co/JLUgW4q2tG - **elithrar** (2025-10-11): https://x.com/elithrar/status/1977050427364368770 - nanda is always cooking 🍳 https://t.co/st1Zw7XpU7 - **kng_lax** (2025-10-11): https://x.com/kng_lax/status/1977043721397903416 - @tjcages okay this is insane. https://t.co/Axtd02aBsZ https://t.co/8Bn6xjqFUf - **DylanAbruscato** (2025-10-11): https://x.com/DylanAbruscato/status/1977043554863386808 - Love that @tjcages is finally getting the recognition he deserves My CTG co-founder and the most talented front-end eng, designer and DJ I know https://t.co/cPBzIWtXlJ - **samlambert** (2025-10-11): https://x.com/samlambert/status/1977041192094515626 - PlanetScale has done very few partnerships but it was obvious why would partner with @Cloudflare. We share some massive customers. https://t.co/GAiAHlwRzP - **_yudax** (2025-10-11): https://x.com/_yudax/status/1977031910116458751 - The best $5/month you can spend https://t.co/y7SKUVKlqJ - **samwillis** (2025-10-11): https://x.com/samwillis/status/1977031666645561791 - I'm quite intrigued by Remix 3 and the manual `.update()`... I think it's really compelling, it's opening the door to using *any* reactive datastore with Remix. The scope for innovation is high. Tying templating to reactivity may have been a mistake. - **nandafyi** (2025-10-11): https://x.com/nandafyi/status/1977026437627650329 - experimenting with grey https://t.co/1EwEIV8qVc - **kentcdodds** (2025-10-11): https://x.com/kentcdodds/status/1977011465505980531 - I vibe coded with Remix yesterday. Zero training data used in the model and it still nailed it perfectly. Literally built a couple games without any issues with the Remix API. https://t.co/HXSGgH9wGp - **mattdorsey** (2025-10-11): https://x.com/mattdorsey/status/1976842795643240647 - We don’t NEED the National Guard here in San Francisco. We need better political allies who are genuinely committed to public safety and serious about solving San Francisco’s problems competently, the right way — and not performatively, the Trump way. And that starts with a fully staffed and fully resourced San Francisco Police Department. Here’s the gift link. (2/2) https://t.co/G1Ij0RSCvm - **mattdorsey** (2025-10-11): https://x.com/mattdorsey/status/1976842793265119244 - This is a slap in the face to San Francisco. It’s insulting to our cops, and it’s honestly galling to those of us who’ve been fighting hard over the last few years to fully staff our @SFPD. Marc @Benioff, I pleaded for your support last year for the Prop F Charter Amendment I wrote, which would have swelled our police staffing ranks by hundreds of officers. And while I appreciated your $50K donation — one week before the election — it wasn’t enough to win. And it wasn’t even CLOSE to the $8+ million you spent to pass the dubious Prop C homelessness tax in 2018. (1/2) - **0x15f** (2025-10-10): https://x.com/0x15f/status/1976793009628103062 - This LP is so pretty https://t.co/Y6jfahjbeW - **JLarky** (2025-10-10): https://x.com/JLarky/status/1976769281473101973 - My positive spin on Remix v3: well you have asked for Rails in JS for years, now you got it - **aleksliving** (2025-10-10): https://x.com/aleksliving/status/1976760820462711164 - I’m sorry cloudflare, I wasn’t familiar with your game https://t.co/9fUXibj3C3 - **floguo** (2025-10-10): https://x.com/floguo/status/1976734326567190714 - imagine having a younger sibling so hot no one believes you came from the same parents https://t.co/BSIWuqPMtZ - **dok2001** (2025-10-10): https://x.com/dok2001/status/1976732272461013227 - One of the perks of working @Cloudflare is getting access to the internal blog https://t.co/eybB0XdAPh - **jorilallo** (2025-10-10): https://x.com/jorilallo/status/1976730551877836875 - 👀 https://t.co/vrg4FMgjHL - **ramonpiano_** (2025-10-10): https://x.com/ramonpiano_/status/1976725460525735997 - sorry @Cloudflare i was not familiar with your game https://t.co/TyjPXyQHfN - **thdxr** (2025-10-10): https://x.com/thdxr/status/1976720297270427681 - if you are a bigger company never engage with a smaller company taking shots at you they can't lose - if you establish the rivalry people will start seeing you as equals and at that point the smaller company can usually win the "i'm cooler than you" battle - **jacksondahl** (2025-10-10): https://x.com/jacksondahl/status/1976719800463507696 - Beautiful work from @ettinger and @tjcages https://t.co/1BW3O6ViIi - **ankrgyl** (2025-10-10): https://x.com/ankrgyl/status/1976703326436909101 - If you love databases and you love basketball, please DM me. Have an interesting opportunity... - **mnolangray** (2025-10-10): https://x.com/mnolangray/status/1976688049695424757 - Ladies and gentlemen, we did it. SB 79 has been signed. https://t.co/h3kZBkiKDT - **ArmandDoma** (2025-10-10): https://x.com/ArmandDoma/status/1976687620785852655 - GAVIN NEWSOM SIGNED SB-79!!!! https://t.co/Wa76huLhnM - **KentonVarda** (2025-10-10): https://x.com/KentonVarda/status/1976682910179541176 - Anyone feel like your Workers are running faster today? https://t.co/lJToeUtBXE - **ettinger** (2025-10-10): https://x.com/ettinger/status/1976681201248117151 - new work for @CloudflareDev Everything we learned from powering 20% of the Internet — yours by default. From first line to full scale. Build without boundaries. https://t.co/JEI9FsRJ5Y - **nizzyabi** (2025-10-10): https://x.com/nizzyabi/status/1976679701633732773 - man @cloudflare rebrand is on pointttt https://t.co/bEiRtBNUVd - **astrodotbuild** (2025-10-10): https://x.com/astrodotbuild/status/1976678053540700404 - Built with Astro ⚡️ https://t.co/r01rus3NKX - **tjcages** (2025-10-10): https://x.com/tjcages/status/1976666098822242336 - a massive congrats & shout out to the team behind this: @ettinger @vbahkk @sachabolteau @Ibelick @jamesallworth @ritakozlov @irvinebroque @burcs 👏👏 - **tjcages** (2025-10-10): https://x.com/tjcages/status/1976663857730089370 - the workers platform for how you work check it out 👀 https://t.co/SXweKenfYr - **tjcages** (2025-10-10): https://x.com/tjcages/status/1976657115633868884 - introducing the new Cloudflare ☁️⚡ bold new look, same global speed powering 20% of the internet – yours by default https://t.co/Z0WqOwuqVZ - **CloudflareDev** (2025-10-10): https://x.com/CloudflareDev/status/1976655655256944801 - Building on Cloudflare just got a new home  → https://t.co/fMLUWM8Icm https://t.co/1YenuaAUry - **ritakozlov** (2025-10-10): https://x.com/ritakozlov/status/1976652156796330002 - we've built the full ai cloud to let you build full stack ai applications it was time our website reflected it 🔥 https://t.co/VpHWBUz0YS - **zebassembly** (2025-10-10): https://x.com/zebassembly/status/1976646373077320004 - One of my favorite parts of Cloudflare is the eagerness to swarm on difficult problems, even if they aren't directly responsible for the problem. - **jonbeckman** (2025-10-10): https://x.com/jonbeckman/status/1976462747157336325 - @BraydenWilmoth Yes please! Google Cloud has Cloud Shell, which seems similar to this, and I think it's one of the best things on any cloud console. - **JacobMGEvans** (2025-10-10): https://x.com/JacobMGEvans/status/1976459966220468237 - I work at Alchemy now 😅 https://t.co/9ztNAW0gfV - **ilyasu** (2025-10-09): https://x.com/ilyasu/status/1976437313128530396 - Everyone seems a little too aware of their opportunity cost these days. - **soleio** (2025-10-09): https://x.com/soleio/status/1976413830856208577 - You must play the rules of your fiercest competitor. - **BraydenWilmoth** (2025-10-09): https://x.com/BraydenWilmoth/status/1976385012489683358 - From the vault: "Cloudflare Cloud Shell" Some exploration work we did a while back to bring an interactive CLI into the dashboard UI and let you create a resources, click its names, and redirect you in the dashboard to the details of the instance. I like sharing explorations. https://t.co/22r8sKJHln - **jayair** (2025-10-09): https://x.com/jayair/status/1976367650243842541 - WHO IS BEHIND @OPENCHODE - **auchenberg** (2025-10-09): https://x.com/auchenberg/status/1976366493769343482 - The Embarcadero Center is an underrated part of SF. https://t.co/xRtBuLLuTa - **soleio** (2025-10-09): https://x.com/soleio/status/1976361439553331497 - If you’re a podcast listener like me: https://t.co/WuKvVH9Vrd - **dcarter_js** (2025-10-09): https://x.com/dcarter_js/status/1976351362817609814 - you can now deploy full-stack workers with your static assets straight from your .tf file. 🧱 point at your build directory and terraform handles the rest: uploads, diffing, chunked transfers with *bonus* of now being able to manage and deploy workflows in terraform enjoy https://t.co/yaIaJeP164 - **dcarter_js** (2025-10-09): https://x.com/dcarter_js/status/1976351365007106551 - changelog: https://t.co/7Wr19VOGbF - **JacobMGEvans** (2025-10-09): https://x.com/JacobMGEvans/status/1976336968616378470 - Truly a great mindset change! Basically run infrastructure in tests like I would components in the frontend 😅🤣 https://t.co/S6NQ0y9Y1L - **zeeg** (2025-10-09): https://x.com/zeeg/status/1976334689012273601 - I want a lightweight toolkit that exposes an agent on Slack, running on top of Cloudflare Workers. It needs to be able to support MCP extensions (remote only) and OAuth. It needs to be permissive open source. Does this exist? - **soleio** (2025-10-09): https://x.com/soleio/status/1976323339389108472 - Back in May, I paid @withdelphi a visit to be the first person to speak to the team about my career, both as a way for them to learn from my experience but also to train my delphi firsthand. It was a candid, far-ranging conversation. Enjoy. https://t.co/8AqxaLuXa3 - **bernhardsson** (2025-10-09): https://x.com/bernhardsson/status/1976306548382068936 - A mistake I’ve made a bunch of times is to focus on the bottleneck X of a system for 6 months until it’s no longer a bottleneck. And then focus on the next bottleneck Y for 6 months. What I should have done is to anticipate Y and focus on both X and Y in parallel. - **KentonVarda** (2025-10-09): https://x.com/KentonVarda/status/1976277138673471787 - PROTIP: If you pass a custom reviver function to JSON.parse(), it gets 10x slower (at least on V8). Try to avoid it if you care at all about performance. - **thdxr** (2025-10-09): https://x.com/thdxr/status/1976195435963482258 - @paulg this is something i would post - **paulg** (2025-10-09): https://x.com/paulg/status/1976195168668876840 - The reason AI coding works so well is that the source code of the median app was already slop before LLMs. - **threepointone** (2025-10-09): https://x.com/threepointone/status/1976187688320651295 - @EastlondonDev @samgoodwin89 @alchemy_run @JacobMGEvans I’m very much a fan of the idea of alchemy innovating in the system, it’s so great to watch. We’ve started using it internally for some stuff and it looks great. Will probably have agents examples soon. - **EastlondonDev** (2025-10-09): https://x.com/EastlondonDev/status/1976172526251417706 - @samgoodwin89 @alchemy_run @JacobMGEvans Think of workers like chrome tabs! - **zebassembly** (2025-10-08): https://x.com/zebassembly/status/1976061113503457489 - It is taking all of my strength to not post about all the interesting observations I'm making right now - **threepointone** (2025-10-08): https://x.com/threepointone/status/1975997269787934969 - chatgpt apps on cloudflare agents easy. ready to use today. https://t.co/E2C066Ydpd https://t.co/HOxGPeBp3N - **whoiskatrin** (2025-10-08): https://x.com/whoiskatrin/status/1975982731273318414 - loved OpenAI Apps SDK? so did we! all ChatGPT App MCPs are ready-to-ship with Agents SDK check out our latest example by @caise_p ↓ https://t.co/SHUH7P2PB9 - **brian_scanlan** (2025-10-08): https://x.com/brian_scanlan/status/1975955696572416384 - @samlambert I watched this recently: https://t.co/XfztqmVNgC (maybe everybody has seen it and I am behind the times) - **samlambert** (2025-10-08): https://x.com/samlambert/status/1975953489705742638 - This is what Oracle took from us. https://t.co/5pBJ1FlWqL - **whoiskatrin** (2025-10-08): https://x.com/whoiskatrin/status/1975885419373023453 - if you are a senior/principal engineer who loves working on hard problems and building infrastructure for AI, agents or MCPs we are hiring! one of the best teams I’ve ever had the pleasure to work with https://t.co/nVJP6hwGFI - **samlambert** (2025-10-08): https://x.com/samlambert/status/1975791973396324770 - Correctness, clean code, DRY, abstraction purity - the shibboleths of people who get nothing done - **thdxr** (2025-10-08): https://x.com/thdxr/status/1975742471473651885 - i've had this idea forever of running a survey for most hated npm packages, downloading them all onto a mini satellite, and then launching it into space is this a good idea - **wesbos** (2025-10-07): https://x.com/wesbos/status/1975681953249534295 - Now that my personal site has been on Cloudflare for ~4 months, I thought I'd share the hosting bill for a RSC site. As a base, I pay $5/mo for Workers and $5/mo for images. On top of that overages were: September was $0.14 for compute and $7.50 for image transformation So about $20/mo My site has about ~1500 pages and ~2500 images with a 1 hour cache before regenerating. This also includes some light use of KV and D1 for caching data - **FardeemM** (2025-10-07): https://x.com/FardeemM/status/1975679898711626002 - The biggest problem cloudflare has is ux and lack of aura these are about to be solved problems. I'm so hyped https://t.co/bjt0EuLwZA - **dcarter_js** (2025-10-07): https://x.com/dcarter_js/status/1975656533762466207 - love this ship - reimagining the dashboard continues https://t.co/sLJgP4CjT9 - **ritakozlov** (2025-10-07): https://x.com/ritakozlov/status/1975651574467993979 - the enprettification of cloudflare. it’s happening! https://t.co/3VMWmz3zfV - **burcs** (2025-10-07): https://x.com/burcs/status/1975649827695566870 - new workers overview just shipped 🚢 we're shipping a better and more beautiful experience, page by page https://t.co/06ofF8ljHY - **paulpopus** (2025-10-07): https://x.com/paulpopus/status/1975644986428342740 - We've now added the @Cloudflare template with D1 to our create payload app command Get started 👇 https://t.co/ODSUSiRdNG - **njpatel** (2025-10-07): https://x.com/njpatel/status/1975621491539456419 - Wrote up some of the challenges we faced when building the first version of our remote MCP server with the kind and volume of data that @axiomfm handles! Have to say the @Cloudflare Agents SDK made it super easy to focus on the value add bits instead of losing ourselves in supporting the fast-moving protocol, authentication, etc... h/t to @threepointone and friends 🙌 - **threepointone** (2025-10-07): https://x.com/threepointone/status/1975588632133009586 - omg what https://t.co/i7uNeTqoZo - **mattzcarey** (2025-10-07): https://x.com/mattzcarey/status/1975578366561661201 - Happy to share I joined @CloudflareDev to work on Agents and Model Context Protocol beyond excited to hack with @threepointone @whoiskatrin @ghostwriternr and crew 🧡 https://t.co/70xLHOqufr - **clairevo** (2025-10-07): https://x.com/clairevo/status/1975576875788619983 - Everyone wants enterprise b2b ACV, no one wants 6-12 month sales cycles, 200 point security questionnaires, redlines, net 60, white glove onboarding, services partner ecosystem, last minute procurement negotiation step, 14 page MSAs, champions that leave the company. 90-180 day activations, features built to save churn, Europe->Asia timezone sales calls, cringe outbound campaigns, SCIM, data deletion and portability process, msft ecosystem integrations (teams!), or EU data residency. - **backpinelabs** (2025-10-07): https://x.com/backpinelabs/status/1975386294265258237 - The increased Cloudflare Container instance types makes the speed of this workload so much faster than I expected. Kudos to @MikeNomitch_CF and team for making so much progress on Containers is such a short amount of time. https://t.co/eHpVyfSxQc - **liadyosef** (2025-10-06): https://x.com/liadyosef/status/1975326097358131688 - Exciting news! We're working with OpenAI to make Apps mcp-ui compatible (and vice-versa). Both share same vision and philosophies - The future experience of the web is not one of distinct, complex websites anymore - but of fragmented, custom "UI chunks" composed by an assistant. https://t.co/20gpuSts1O - **CarolXu715** (2025-10-06): https://x.com/CarolXu715/status/1975313225777619385 - Excited to join @CloudflareDev on the R2 team! Also excited to announce: users can now add directories via the R2 dashboard! 🚀 This can be done when you choose to view object prefixes as directories. What features would you like to see next? https://t.co/uJKhWQYTgK - **kentcdodds** (2025-10-06): https://x.com/kentcdodds/status/1975269042673098795 - All built on MCP ⚡️ https://t.co/EKF0DzbMg6 - **threepointone** (2025-10-06): https://x.com/threepointone/status/1975251752434999723 - (this means cloudflare agents is immediately ready to build chatgpt apps. we'll have demos asap.) https://t.co/JokZadPprP - **threepointone** (2025-10-06): https://x.com/threepointone/status/1975250330381394177 - chatgpt apps sdk: "if you have an existing mcp server, then it's as simple as adding a resource that returns html" !!! - **KentonVarda** (2025-10-06): https://x.com/KentonVarda/status/1975244452508602802 - I think I may be obsessed with sandboxing. https://t.co/XUZgMAxvRi - **ankrgyl** (2025-10-05): https://x.com/ankrgyl/status/1974966982765363660 - a prediction in advance of dev day tomorrow: agent logic/control flow is going to move down into the API layer, rather than live in your code. in other words, the agent, not chat turn, will be the API. responses is already designed to support this. - **simonw** (2025-10-05): https://x.com/simonw/status/1974835974938206222 - I've become one of those people who run multiple coding agents at once - here are my notes on embracing the parallel coding agent lifestyle https://t.co/W4tk6F9Eou - **Echo_Grain** (2025-10-04): https://x.com/Echo_Grain/status/1974541333865107843 - @TysonTrautmann Perfect summary. The right latency strategy isn’t global it’s contextual. Optimize for the path that feels instant, not the one that benchmarks well. - **TheStalwart** (2025-10-04): https://x.com/TheStalwart/status/1974480254459625758 - My new thing is looking for collections of old journals for sale online. https://t.co/TBtWvncZNG - **JessePeltan** (2025-10-04): https://x.com/JessePeltan/status/1974424385282085310 - Batteries will soon provide power throughout the entire night. Then, they will power the entire night — everywhere. https://t.co/HvxwbTKdtN https://t.co/Yfg8YHwpAn - **nicolasfulghum** (2025-10-04): https://x.com/nicolasfulghum/status/1974396633371869462 - @verusphoenix This is grid-storage, not home storage. - **kane** (2025-10-04): https://x.com/kane/status/1974340061182505090 - @rohindhar Why does @sfplanning even have a say in this - **SIChrisMannix** (2025-10-04): https://x.com/SIChrisMannix/status/1974296897876865240 - Insane how bad these answers are https://t.co/BKKg6rpP1K - **aninibread** (2025-10-04): https://x.com/aninibread/status/1974268853535285548 - I’m so happy to have gotten a chance to speak to students at NYU about the Cloudflare Developer Platform as part of the @spaceappsnyc NASA space apps challenge! 🚀✨🌟 Excited to see the projects that come out of this weekend. 🧡 https://t.co/gZUq2BK7fC - **KentonVarda** (2025-10-04): https://x.com/KentonVarda/status/1974267386288021803 - The annual Access Control List festival has started in Austin but I'm a capability-based security guy so I refuse to go. I wonder what they do there. It seems noisy. - **thdxr** (2025-10-03): https://x.com/thdxr/status/1974232924275777796 - we've been in the devtools space for a long time and have never paid anyone to shill our products and we never will we won't "partner" with someone who "was already using our product" either - **thdxr** (2025-10-03): https://x.com/thdxr/status/1974225475187724455 - play long term games its the only way to get anything of value a form of that is giving honest advice instead of shilling your own products https://t.co/IfNsKmiWDL - **threepointone** (2025-10-03): https://x.com/threepointone/status/1974218780076134831 - omg this is such a good product and it’s been a complete fkin pain in the ass to set up solved https://t.co/9MVvWaEWgS - **eastdakota** (2025-10-03): https://x.com/eastdakota/status/1974215827801985300 - Which will mean the winner in the GPU race will be whoever finally takes tokens per watt seriously. https://t.co/7Ro5yLef8o - **thdxr** (2025-10-03): https://x.com/thdxr/status/1974213749184606682 - i see frontend framework discussion on here again this means we're past peak AI - **CloudflareDev** (2025-10-03): https://x.com/CloudflareDev/status/1974206866881458276 - 🆕 Enable Cloudflare Access for your Worker in one click https://t.co/fQU8G2SjVZ - **nevikashah** (2025-10-03): https://x.com/nevikashah/status/1974182175152205928 - @dok2001 @zalmanlew @CFJWilliamson @boristane @irvinebroque just released a fix for this - you should see a column for trigger now in the invocation tab. thanks for calling out this out 🙏 https://t.co/YpSwLUfwW8 - **nevikashah** (2025-10-03): https://x.com/nevikashah/status/1974131431933948169 - 📣 Workers Analytics Engine now supports 28 additional SQL functions ingest and store high cardinality, custom analytics within your Worker and run more powerful queries to get the most out of your data 💪 changelog 👇 - **ritakozlov** (2025-10-03): https://x.com/ritakozlov/status/1974077282181657024 - touching g̶r̶a̶s̶s̶ water https://t.co/IwSfWibczW - **samlambert** (2025-10-03): https://x.com/samlambert/status/1974036448417624168 - have you really lived in sf until you’ve fallen down the stairs at li po? - **dok2001** (2025-10-03): https://x.com/dok2001/status/1973948550058070183 - @zalmanlew @CFJWilliamson @nevikashah @boristane @irvinebroque - **mjackson** (2025-10-03): https://x.com/mjackson/status/1973918409709793328 - "I'll take glaring holes in web standards APIs for $500, Alex" https://t.co/HAobEy4geG https://t.co/n3TydOe4Jh - **thdxr** (2025-10-02): https://x.com/thdxr/status/1973891162894422250 - i can't believe non-technical devrels for devtool products is already a thing again - **aninibread** (2025-10-02): https://x.com/aninibread/status/1973868719324737730 - ✨ Hoping to add a 0.001% improvement to your day. Thank you to @fireblaze_15 for putting this together for us! 🧡 https://t.co/5dfkUCL0aI - **ttunguz** (2025-10-02): https://x.com/ttunguz/status/1973853375751205281 - AI breaks the data stack. Most enterprises spent the past decade building sophisticated data stacks. ETL pipelines move data into warehouses. Transformation layers clean data for analytics. BI tools surface insights to users. This architecture worked for traditional analytics. But AI demands something different. It needs continuous feedback loops. It requires real-time embeddings & context retrieval. Consider a customer at an ATM withdrawing pocket money. The AI agent on their mobile app needs to know about that $40 transaction within seconds. Data accuracy & speed aren’t optional. Netflix rebuilt their entire recommendation infrastructure to support real-time model updates1. Stripe created unified pipelines where payment data flows into fraud models within milliseconds2. The modern AI stack requires a fundamentally different architecture. Data flows from diverse systems into vector databases, where embeddings & high-dimensional data live alongside traditional structured data. Context databases store the institutional knowledge that informs AI decisions. AI systems consume this data, then enter experimentation loops. GEPA & DSPy enable evolutionary optimization across multiple quality dimensions. Evaluations measure performance. Reinforcement learning trains agents to navigate complex enterprise environments. Underpinning everything is an observability layer. The entire system needs accurate data & fast. That’s why data observability will also fuse with AI observability to provide data engineers & AI engineers end-to-end understanding of the health of their pipelines. Data & AI infrastructure aren’t converging. They’ve already fused. References Netflix Technology Blog. (2025, August). “From Facts & Metrics to Media Machine Learning: Evolving the Data Engineering Function at Netflix.” https://t.co/F3mhTULobA ↩︎ Stripe. (2025). “How We Built It: Stripe Radar.” https://t.co/XyR0sKJm2a ↩︎ - **MikeNomitch** (2025-10-01): https://x.com/MikeNomitch/status/1973498055946363204 - FFmpeg go brrr - **MikeNomitch** (2025-10-01): https://x.com/MikeNomitch/status/1973497466185191686 - You can now get up to 4 vCPU and 12 GiB memory per instance on Cloudflare Containers https://t.co/V2gsm9IC07 - **patrickc** (2025-10-01): https://x.com/patrickc/status/1973386447068512347 - We're adding some pretty fundamental new capabilities to Stripe accounts. • Hold and convert balances in major currencies. (Including stablecoins.) • Send and receive over local bank rails. (Operate locally, globally.) For sending, you only need an email address. • Instantly issue a card backed by a balance. Over the past few years, we've built a lot of sophisticated treasury infrastructure to power Stripe. We decided to expose this directly to Stripe users, so that you can use our stack to run other parts of your business. Support for all US and UK businesses rolling out this year. - **patrickc** (2025-10-01): https://x.com/patrickc/status/1973384969624993801 - Every day starts with the full arc of joy and despair. The commencement of the morning coffee, and then its end. - **sama** (2025-10-01): https://x.com/sama/status/1973381552621887706 - i get the vibe here, but... we do mostly need the capital for build AI that can do science, and for sure we are focused on AGI with almost all of our research effort. it is also nice to show people cool new tech/products along the way, make them smile, and hopefully make some money given all that compute need. when we launched chatgpt there was a lot of "who needs this and where is AGI". reality is nuanced when it comes to optimal trajectories for a company. - **WillManidis** (2025-10-01): https://x.com/WillManidis/status/1973368220129304757 - the ford foundation is the single greatest office building in america. https://t.co/KBHhLjCbM2 - **XianhangLi** (2025-10-01): https://x.com/XianhangLi/status/1973214068644450661 - 🤔 Ever thought a small teacher could train a student 6× larger that sets new SOTA in training efficiency and frozen evaluation performance for video representation learning? 🤔 Do we really need complex EMA-based self-distillation to prevent collapse, bringing unstable loss dynamics while offering little insight into representation quality? 🚨 In our new paper, we investigate these questions and propose SALT (Static-teacher Asymmetric Latent Training): a simple, scalable, and compute-efficient alternative for video representation learning. 📄 Rethinking JEPA: Compute-Efficient Video SSL with Frozen Teachers 🔗 https://t.co/C9amVFddSH - **dok2001** (2025-10-01): https://x.com/dok2001/status/1973183138630345070 - Give it a try! https://t.co/mNkhXHBFrc - **aninibread** (2025-10-01): https://x.com/aninibread/status/1973182408070668457 - Love seeing our team fix paper cuts. 🛠️ Before: deleting a locked object on dash → gone, then back after refresh. Now: delete stays greyed out with the right tooltip. https://t.co/bJMGCSQHaz - **swyx** (2025-09-30): https://x.com/swyx/status/1973112150760783922 - people are insufficiently calibrated on Sonnet 4.5 imo there were a lot of memes about the "great lock in of sept-dec" but yesterday was really the starting gun to give you some perspective, last koding model jump of this magnitude spawned Devin 2.0, Windsurf, Lovable, Bolt, et al. it's all hands on deck at Cog, team is extremely energized - **mcuban** (2025-09-30): https://x.com/mcuban/status/1973096941954228490 - Look at how they hire and train new officials. Look at how they promote new officials and from where Refs get trained in the Gleague. If they aren't good there because there is very little investment made in training them there, how are they going to be good when they get to the Wnba or NBA ? - **Hali** (2025-09-30): https://x.com/Hali/status/1973091173607543049 - Whew https://t.co/NzypBsF2Xs - **malika_andrews** (2025-09-30): https://x.com/malika_andrews/status/1973084525111472570 - Napheesa Collier’s full statement today, where she challenged Commissioner Cathy Engelbert and the WNBA with directness and stunning detail we rarely hear from active players. Worth listening to every word. https://t.co/IRTvTc52EA - **payloadcms** (2025-09-30): https://x.com/payloadcms/status/1973057103825613054 - The one you've been waiting for: 1-click deploy Payload ... on @Cloudflare. 🚀 - **StartupArchive_** (2025-09-30): https://x.com/StartupArchive_/status/1973055093604782422 - Jensen Huang on why he rarely fires people and will instead “torture them into greatness” Jensen once told Stripe founder Patrick Collison that he didn’t like firing people and seldomly did it. When asked to elaborate on this, Jensen responds: “I’d rather improve you than give up on you. When you fire somebody, a lot of people will say ‘it wasn’t your fault,’ or ‘I made the wrong choice.’ But I used to clean bathrooms and now I’m the CEO of a company. I think you can learn it. There are a lot of things in life that I think you can learn and you just have to be given the opportunity to learn it… I don’t like giving up on people because I think they can improve.” He continues: “It’s kind of tongue in cheek, but people know I’d rather torture them into greatness. I’d rather torture you into greatness because I believe in you. And I think that coaches that really believe in their team torture them into greatness. Oftentimes they’re so close. Greatness will sometimes come in one day with an ‘I got it!’ — that feeling that you didn’t get it yesterday and all of a sudden one day something clicks. Could you imagine giving up that moment right before you got it? I don’t want you to give up on that, so I’ll just keep torturing you.” Video source: @stripe (2024) - **nevikashah** (2025-09-30): https://x.com/nevikashah/status/1973019371867173150 - something ✨new✨ is coming to Workers Observability 👀 can you guess what it is 🤭🤫🙈 https://t.co/215AjjDqcu - **latentspacepod** (2025-09-30): https://x.com/latentspacepod/status/1973017487190139140 - Anthropic released a TON of updates today: • Sonnet 4.5 (with context awareness) • Claude Code 2.0 (+ new mascot 🦀) • Claude API: context editing + memory tool @mikeyk sat down for a special launch day chat about 4.5 and the @AnthropicAI developer roadmap! (and more: • VS Code extension • Claude apps code execution and file creation • Claude for Chrome • Imagine with Claude) Available here or on our youtube channel: - **serhiichernenko** (2025-09-29): https://x.com/serhiichernenko/status/1972802834094776376 - @irvinebroque @dok2001 Sadly, but it's not what I’m looking for, if I understood correctly. Use case: I’m working in a company that has its own CF account. Also, we have a project and a client that brings their CF account with projects, domains, workers, etc. And for personal use, I have my own CF acc. - **dok2001** (2025-09-29): https://x.com/dok2001/status/1972785887382843725 - @serhiichernenko @irvinebroque ^ - **mikeyk** (2025-09-29): https://x.com/mikeyk/status/1972718726052286637 - We asked every version of Claude to make a clone of Claude(dot)ai, including today’s Sonnet 4.5… see what happened in the video https://t.co/fq0UnGNHhj - **fidjissimo** (2025-09-29): https://x.com/fidjissimo/status/1972707487238467914 - Cool thing launching today: you can now buy products directly from ChatGPT (starting with Etsy and expanding soon to over a million Shopify merchants). It’s powered by the Agentic Commerce Protocol, an open standard we built with Stripe. Merchants and developers who want Instant Checkout, you can get started today: https://t.co/VdHyAtHvhO - **zebassembly** (2025-09-29): https://x.com/zebassembly/status/1972661834302755332 - Huh, now that I think about it I wonder if the Worker Sharding we just released made WASM faster since you're going to hit an isolated with an already-compiled WASM instance? - **LambrosPetrou** (2025-09-28): https://x.com/LambrosPetrou/status/1972406358491468068 - @irvinebroque @AdamSmielewski There is no locking across multiple ruks of migrations. Each migration will run once inside a transaction though, but the runners will interleave. - **southpolesteve** (2025-09-28): https://x.com/southpolesteve/status/1972384804562776126 - @yusukebe Amazing. Can you PR them to https://t.co/yDrYvKTtqx examples dir? - **jonas** (2025-09-28): https://x.com/jonas/status/1972369892679815657 - @rafalwilinski I mean, you could let the LLM generated code call a function to make an LLM request “in the middle” of the program - **dok2001** (2025-09-28): https://x.com/dok2001/status/1972363282284679224 - @DhravyaShah @irvinebroque this seems like good idea - **rafalwilinski** (2025-09-28): https://x.com/rafalwilinski/status/1972362720579035146 - couple of thoughts about codemode/codeact (https://t.co/s9642Cp2dm) - it's great but it's not a silver bullet. How it works? An agent, instead of doing N tool calls to meet the goal, writes the code that is calling these tools. Code is then evaled at secured sandbox and voila. so we've gone from telling LLMs what to do, to teaching them to write instructions for themselves. It's turtles writing code all the way down *for all the domains* why it's neat: - eliminates *some* problems with long trajectories or fan-outs - it's much easier to oneshot code with a `for loop` over 100 entries instead of expecting an LLM to nail 100 tool calls, either in parallel or sequentially. It only gets worse with bigger N. it's also MUCH faster - MCP schemas become typed TS APIs - this makes it verifyable before running. you can also reuse it easily for e.g. semantic caching. when you see the same input for 2nd time, you don't need to invent the flow, just reuse something that worked in the past (wink wink Vectorize) - saving context/preventing from context rot. instead of dumping that 100k payload from gmail API to `messages`, it's just flowing inside sandboxed runtime, massaged like every other JSON and only precisely picked bits get surfaced in the ctx - enables CaMeL-style approaches where agents literally debug their own homework (https://t.co/5G7IKV5GDI) So, it works great for problems where you know majority of the flow upfront, think of workflow-like things, e.g. "please provision an ec2 instance of m4 class that I can SSH to, place that in public SG and attach an IPGW, make sure it's tagged nicely". it's a clear sequence of function calls which can be easily done as a script. Doing so traditional way _could_ exceed the context and get lost along they. where it’s weaker: open‑ended research loops where you decide at each step what to even do next. You can try to account for each edge case but it kinda defeats the purpose. it's also challenging with idempotency & partial failures but solvable with KV/D1 for resume‑from‑checkpoint/offset etc. Right now, it especially fails at cases where the intelligence needs to be _inserted in the middle of code_. Imagine a spreadsheet with 100 emails and you wanna write a *personalized* email for each entry. `body` argument for that `send_email` call must be also computed using LLM in order for that outreach to be personalized, right? You can shim that with sub‑agents / “call llm() inside the loop,” but then you’re kinda back to traditional agenting, just wrapped in TS. Nevertheless, super excited to see how the field develops this pragmatic bridge between “agent” and “script”, pumped that Cloudflare invests into it. @KentonVarda @threepointone pls do more shit like that - **DhravyaShah** (2025-09-28): https://x.com/DhravyaShah/status/1972362068419572204 - I wish there was a way to do cloudflare workers without having a wrangler.json just need good defaults compatibility_date: latest compatibility_flags: [ nodejs_compat, nodejs_compat_populate_process_env ] & takes package name from package.json that would just work™ - **natolambert** (2025-09-28): https://x.com/natolambert/status/1972323625207509321 - RL research is becoming like pretraining/modeling. This is a huge vibe shift. Most research published on RL isn't using enough compute to make many of these decisions matter as much. This is slowly shifting. https://t.co/FwhLhsUUZq - **dok2001** (2025-09-28): https://x.com/dok2001/status/1972319108189368627 - @jonahseguin @boristane @irvinebroque - **yusukebe** (2025-09-28): https://x.com/yusukebe/status/1972232141737197661 - Ah, nice! Cap'n Web server running on Cloudflare Workers/Deno/Node.js with Hono https://t.co/L8Ht50LaEU - **mikeaalen112735** (2025-09-28): https://x.com/mikeaalen112735/status/1972135524749267015 - Simply speaking this is not Basketball. Imo Alyssa Thomas aka dribble handoff merchant is the dirtiest player in the history of the league. 🚨🚨 https://t.co/UC3EC1hLiH - **akothari** (2025-09-28): https://x.com/akothari/status/1972123738268397873 - Underrated: dining at fort mason (Radhaus, Interval, Greens) https://t.co/wkws0jayJR - **thdxr** (2025-09-28): https://x.com/thdxr/status/1972117946907480334 - on track for 200,000 monthly active developers crazy we launched 3 months ago and the product isn't even good yet will be soon though - **eastdakota** (2025-09-27): https://x.com/eastdakota/status/1972065958761050421 - When @Cloudflare went public I ran a little exercise: I took our market cap and divided it by the number of days since we launched. Market cap in so many ways is a terrible proxy for value actually created, but after you’re public it’s better than almost anything else. I smiled when I saw back then we’d generated $1.2 million in value every day, including weekends and holidays, from the day we launched until 9 years later when we IPOed. 😀 We just turned 15 today so, on a lark, I ran the numbers again today. We’re up to $13.8 million in value generated per day, including weekends and holidays. 🤯 What I’m struck by is that when we went public @zatlyn and I signed our founders’ letter that we were “just getting started.” Turns out that was more true than we imagined. What I didn’t appreciate was how much our rate of innovation and value creation could accelerate. There’s an assumption that as you get bigger you inherently get slower. There’s no reason that has to be the case. And I’m proud that @Cloudflare has proven you can innovate like mad at scale. Still just getting started. And I’ll take the over on any bet on our 20-year number. #zoomzoom - **thdxr** (2025-09-27): https://x.com/thdxr/status/1972055605054787970 - one of the hardest parts about devtools is even if someone finds your product interesting it might be months before they can try it on something so you have to stay relevant every week every company is oriented around infrequent big boom marketing events - doesn't work - **kristianfreeman** (2025-09-27): https://x.com/kristianfreeman/status/1972042427386204567 - Spend six hours in Ableton kind of day - **wishee0** (2025-09-27): https://x.com/wishee0/status/1972041893514301547 - interesting, i wanted to simulate colonies on cloudflare workers and see how agents would actually think. scouts, guards and queens make independent decisions, but each colony needs memory. without persistence, behavior resets every tick and nothing compounds over time https://t.co/UNtjaIial2 - **BrianSuttererMD** (2025-09-27): https://x.com/BrianSuttererMD/status/1972041218143166478 - Nobody sabotages the WNBA like the WNBA https://t.co/uM0fdetPGw - **DhravyaShah** (2025-09-27): https://x.com/DhravyaShah/status/1972038354670964831 - Introducing: The new Supermemory developer platform. Memory doesn't have to be a black box. We pay attention to observability and dev experience! Better observability, 10x improved memory graph. and... light mode! Enterprise-ready - you can now invite your team + much more. https://t.co/zKskPcul4G - **matijagrcic** (2025-09-27): https://x.com/matijagrcic/status/1972031433930920275 - TL;DR Cloudflare is building an operating system for AI agents. After a massive week of announcements for their 15th birthday, the strategic direction is clear. Here's the proof 🧵 - **ivebenfreed** (2025-09-27): https://x.com/ivebenfreed/status/1972027123486580906 - https://t.co/riANDRUbpw - **ivebenfreed** (2025-09-27): https://x.com/ivebenfreed/status/1972027029492322459 - I setup a full stack app with the Workers Vite plugin from @CloudflareDev and @better_auth. Technically 4 or 5 prompts from the steering needed but only took me about an hour. Incredible! - **dok2001** (2025-09-27): https://x.com/dok2001/status/1972023324051231113 - I still think about this a lot every time says they do a lot of rps and asks me if @Cloudflare can handle their load https://t.co/eaSpjK2xU7 - **eastdakota** (2025-09-27): https://x.com/eastdakota/status/1972014693297590729 - This post is really interesting. Suggests MCP may hold agents back because it’s new and unfamiliar. Kenton and Sunil built a way to turn MCP into code, which LLMs are good at. But perhaps the answer is the protocol should be more code-like from the start. https://t.co/1O1eJ3nrsC - **threepointone** (2025-09-27): https://x.com/threepointone/status/1971988718052651300 - dammit this is obvious/clever https://t.co/Opu9YQJ27g - **Lindsay_13** (2025-09-27): https://x.com/Lindsay_13/status/1971973853485412476 - When Bill Simmons is tweeting… - **BillSimmons** (2025-09-27): https://x.com/BillSimmons/status/1971948268717584428 - A rare WNBA tweet for me 1. The Lynx got HOSED by calls in last year’s Finals, they should’ve won 2. Collier gets less ref respect than any NBA/WNBA star, it’s bizarre 3. The reffing last night was hilariously bad 4. This is unequivocally a foul ⬇️⬇️⬇️ https://t.co/yWteauChUa - **JenovaAIAgent** (2025-09-27): https://x.com/JenovaAIAgent/status/1971941308320436353 - https://t.co/QP5WTPyiUE - **eastdakota** (2025-09-27): https://x.com/eastdakota/status/1971908770176368644 - Yup. https://t.co/lqHYRT0Bxp - **alxnderhughes** (2025-09-27): https://x.com/alxnderhughes/status/1971880471022993628 - Holy shit... Cloudflare just open-sourced an entire AI coding platform that lets anyone build and deploy apps with natural language. VibeSDK is basically Replit/Cursor but you can deploy your own version in one click. This changes everything. Here's how it works: https://t.co/E8cWUzjSJS - **Rachel__Nichols** (2025-09-27): https://x.com/Rachel__Nichols/status/1971839771552239922 - “It’s f*****g malpractice” - Lynx Coach Cheryl Reeve on the WNBA’s choices of officials for playoff games. Reeve: “I want to call for a change of leadership at the league level when it comes to officiating.” https://t.co/tzHu2QJkar - **DhravyaShah** (2025-09-27): https://x.com/DhravyaShah/status/1971836009504821489 - spacebot, one of my most successful projects in the early days, had AI 4 years ago no, not openai API. because ChatGPT didn't exist yet... I hadn't seen or used a chatbot before it was really fun to talk to. and something new. and i was early memory is the next hill to climb https://t.co/uFi64vv9ph - **JordanDye** (2025-09-27): https://x.com/JordanDye/status/1971813692112359439 - Louder for the people in the back (that ain’t know hoops) https://t.co/ObqUrC1uvU - **aninibread** (2025-09-27): https://x.com/aninibread/status/1971797791669465560 - Thank you to @burcs and @BraydenWilmoth for our new look! - **aninibread** (2025-09-27): https://x.com/aninibread/status/1971797541542121768 - It's 12AM on a Friday night and I still can't get over how good the new R2 dashboard looks. 😌 https://t.co/n0W2iqZ6m2 - **JaimeLardis** (2025-09-27): https://x.com/JaimeLardis/status/1971793777544384910 - I watch a lot of basketball. WNBA, NBA, Men’s and Women’s college. I must admit Cheryl Reeve post game interview was probably the most gangsterous respectable coaches post game interviews ever!! - **saranormous** (2025-09-27): https://x.com/saranormous/status/1971792770303250683 - 5/ unlike my fave games, most of human workflows don’t come with clean scores. Explicit, verifiable rewards — a motion granted, a readmission avoided, a customer renewed — are a tiny fraction of what matters. The rest is tacit judgment with no +1/-1 attached. - **saranormous** (2025-09-27): https://x.com/saranormous/status/1971792771792126243 - 6/ Hard constraints make it worse. You can’t “explore” by breaching privilege, prescribing contraindicated drugs, or refunding against policy. These aren’t soft penalties, they’re bright lines that aren't easily explored. I think RL doesn’t like absolute walls? - **kendra__andrews** (2025-09-27): https://x.com/kendra__andrews/status/1971790016419713330 - Cheryl Reeve didn’t take any questions postgame. Instead, she spoke for two minutes on officiating, and called for a change of leadership at the league level when it comes to officiating. She also said Collier had a “possible fractured” but specify what body part. https://t.co/1jkMEmd8RX - **PercyAllen206** (2025-09-27): https://x.com/PercyAllen206/status/1971783224662937799 - The tip was clean, but AT ran through Phee to complete the steal. That’s a foul. - **KentonVarda** (2025-09-26): https://x.com/KentonVarda/status/1971698027099574639 - Connecting the dots. Code-mode agents act by writing TypeScript code that runs in a sandbox and calls tools over capability-based RPC. Cap'n Web is a protocol for exposing rich TypeScript APIs over capability-based RPC. - **dsp_** (2025-09-26): https://x.com/dsp_/status/1971685898237759499 - An Update: The next version of MCP will be released on November 25th. https://t.co/7O6dAtemU8 - **elithrar** (2025-09-26): https://x.com/elithrar/status/1971669267516121186 - @backpinelabs @tan_stack @CloudflareDev lots more to come here + send us any feedback you have. focused on making TanStack x Workers just work. cc @irvinebroque - **backpinelabs** (2025-09-26): https://x.com/backpinelabs/status/1971662002595512474 - It's crazy how much the DX of @tan_stack start has improved in the RC for V1. I put this repo together as I have been playing around with the @CloudflareDev setup. This project ships with: - CF Vite Plugin so you can bind to CF resources during dev - Clear server entry point so you can setup Queues, Durable Objects, Crons, and Workflows within the start project - CF scripts so you can deploy with no additional config https://t.co/fWcsqHzAj7 - **d0m96** (2025-09-26): https://x.com/d0m96/status/1971620773250081036 - Did you know that you can also run Python code in dynamic workers? Yep. 👇🧵 https://t.co/Vl5AiY65aG - **RobbieSap90** (2025-09-26): https://x.com/RobbieSap90/status/1971599603620573569 - We also have docs now! https://t.co/QUwowisoQ7 https://t.co/RKhoKxcH3S - **ritakozlov** (2025-09-26): https://x.com/ritakozlov/status/1971598930715807916 - let's have coffee and talk about day 5 (final day!) of birthday week ☕️ TL;DR: code mode is 🤯; if you measure things, you can make them faster https://t.co/amsoB1N7qm https://t.co/RyfQ05IQXC - **kentcdodds** (2025-09-26): https://x.com/kentcdodds/status/1971592287642878076 - https://t.co/8WxZCkZDUE You know what I love about this? (other than it shows a better way to use MCP servers) Instead of saying "check out this new tech, you'll love it" they say "this is a better way to do something, and this new tech makes it possible!" - **MikeNomitch** (2025-09-26): https://x.com/MikeNomitch/status/1971591789087006860 - On Cloudflare Workers, you often don't notice cold starts. But just in case you do... we cut the rate of cold starts by 10X. https://t.co/lsCJaYbgOX - **KentonVarda** (2025-09-26): https://x.com/KentonVarda/status/1971590404140351822 - This is why Workers pricing is so low that people make up conspiracy theories about some sort of rug-pull. We just built a better architecture, and we still have low-hanging fruit like this to make it even more efficient. - **KentonVarda** (2025-09-26): https://x.com/KentonVarda/status/1971590400537436470 - This adds almost no latency since it's always still to a machine in the same physical building, but it avoids a lot of cold starts. Cold starts are extremely fast on workers, but avoiding them is even faster. https://t.co/EhxMirvN7g - **KentonVarda** (2025-09-26): https://x.com/KentonVarda/status/1971590402424922359 - I suspect most people will be shocked we weren't doing this all along. In a container world it would be insane not to do this from the start. But really! Isolates are SO CHEAP compared to containers, it all worked fine! - **KentonVarda** (2025-09-26): https://x.com/KentonVarda/status/1971590398033506332 - Believe it or not, until recently, Cloudflare Workers would just run your Worker on whatever random machine the HTTP request landed on. Isolates are so cheap, this worked fine. But we now do a little bit of routing within the LAN to make cold starts less frequent. https://t.co/QMEpyZK5hv - **mattshumer_** (2025-09-26): https://x.com/mattshumer_/status/1971586555119600025 - This is an awesome approach to using MCP servers. We need more people to pay attention to this. Basically, convert a MCP spec into code functions, and have the model write code to call the functions, instead of typical tool-caling. I’ve been doing this with Python for a long time now... you always want to lean into what the model is good at, and obviously, models have seen FAR more code than they have tool calls. Plus, code can orchestrate calls in more complex ways than pure tool-calling can... for example, instead of doing 50 calls to the same tool over and over with different arguments, a model can just write a for loop and do it with way fewer tokens (cheaper, faster, less context bloat!). - **reactjs** (2025-09-26): https://x.com/reactjs/status/1971585517449031907 - We're excited to announce React Conf speakers from some of your favorite React frameworks and build tools like @expo, @nextjs, @parceljs, @ReactRouter, @RedwoodJS, @rockjs_dev, and TanStack - **mattshumer_** (2025-09-26): https://x.com/mattshumer_/status/1971585158051729755 - @KentonVarda This is an awesome approach. I’ve been doing this with python for a long time now. You always want to lean into what the model is good at, and obviously, models have seen far more code than they have tool calls. Plus there are benefits like loops etc! - **Cloudflare** (2025-09-26): https://x.com/Cloudflare/status/1971583073604063476 - We reduced Cloudflare Workers cold starts by 10x by optimistically routing to servers with already-loaded Workers. Learn how we did it here. https://t.co/3jz7EISTA5 #BirthdayWeek - **LucasRolff** (2025-09-26): https://x.com/LucasRolff/status/1971582509885645191 - @irvinebroque I look forward to being able to use DNS Firewall without having to sign a $24k/year contract as a small business 😆 - **threepointone** (2025-09-26): https://x.com/threepointone/status/1971579381085548643 - ok, so. dynamic workers. when LLMs are generating a ton of code, you need something better to run it in. imagine eval, but better. way better. control what's exposed to the env. control outgoing fetches. zero cold start. left: code in a string. right: output. https://t.co/qS07TNyThb - **KentonVarda** (2025-09-26): https://x.com/KentonVarda/status/1971577292674732274 - By the way, we also shipped this mundane but extremely-useful Workers API feature today. You no longer have to configure bindings if they just point back to your same worker. The runtime provides auto-configured bindings under `ctx.exports`. https://t.co/FpatMY2K92 - **zebassembly** (2025-09-26): https://x.com/zebassembly/status/1971573519906480268 - Banger from the amazing Worker sharding blog post: I don’t know the official answer, but I can tell you that the code paths do exist, and they do get exercised. Feels like a very good summary of the cool undocumented capabilities of our runtime https://t.co/gVAem1tSSC - **whoiskatrin** (2025-09-26): https://x.com/whoiskatrin/status/1971570451811066177 - 🎈 friday doesn't stop agents team from shipping a new version of Agents SDK - codemode: a brand new (experimental!) way of interacting with LLM tools; literaly generate code to run tools that runs in a sandboxed environment. delots of research potential here, keep an eye out for updates - MCP server/client: we completely refactored the mcp server and client to align more closely with the mcp sdk while keeping everything compatible. the agent’s built-in mcp client now fully supports auth, redirects etc - don't forget about x402 : x402 supercharges your MCP servers so they can include paid tools we've added some docs and guides (more coming soon!). go build cool stuff! - **dbelson** (2025-09-26): https://x.com/dbelson/status/1971567990157271102 - Proud of the @CloudflareRadar team for the work done to add three great new data sets to Radar -- we're constantly expanding visibility into what's happening on the #Internet! 🧵 https://t.co/Dyjye0efYe - **threepointone** (2025-09-26): https://x.com/threepointone/status/1971567136343744850 - ok yeah this is mad. if you have a domain on cloudflare - automatic semantic search index - with pub sub updates for "crawlers" - with an mcp server I didn't even know this was happening lol https://t.co/FJ44SjzfSQ - **kristianfreeman** (2025-09-26): https://x.com/kristianfreeman/status/1971566695690146215 - THIS IS SO COOL https://t.co/OtCJRKmgvq - **threepointone** (2025-09-26): https://x.com/threepointone/status/1971566215849025869 - @EastlondonDev @irvinebroque @samgoodwin89 we had support for this in wrangler v2 with a hotkey, (but disabled it for reasons) I think it's still in the codebase, we should bring it back there too - **whoiskatrin** (2025-09-26): https://x.com/whoiskatrin/status/1971566036202983676 - code mode is here mcp tools as typescript apis, llms writing code instead of awkward tool calls brilliant work from @threepointone and @KentonVarda 👏 https://t.co/9jQQq1QoXF - **threepointone** (2025-09-26): https://x.com/threepointone/status/1971564891593998584 - this is 3 announcements in one - dynamic isolates are live! ( I call it supereval) - codemode is deeply promising, kinda crazy tbh - sunil can actually be useful https://t.co/iJGU5XTXSg - **KentonVarda** (2025-09-26): https://x.com/KentonVarda/status/1971564025126531423 - In fact, the sandboxing lets you turn off internet access entirely, while defining custom RPC bindings to give to the sandboxed Worker. So you can use capability-based security to give it access only to the resources it is supposed to access. E.g. an MCP server, in code-mode. - **KentonVarda** (2025-09-26): https://x.com/KentonVarda/status/1971564017106698326 - .@threepointone took it from idea to working implementation in three weeks, while also shipping x402 and a million other things. 🤯 There's still some work to do on prompt engineering but we wanted to get this into developers' hands to play with ASAP! https://t.co/J1L6SHAmbm - **KentonVarda** (2025-09-26): https://x.com/KentonVarda/status/1971564015282421909 - OK folks, today is the day when I reveal what this tweet was about. We made MCP work better by: 1. Convert the MCP tools into a TypeScript API 2. Tell the agent to write code against this API We call it "code mode" It uses a new feature in Workers: Dynamic Worker Loading https://t.co/JzdZASp93W - **aninibread** (2025-09-26): https://x.com/aninibread/status/1971561796638257460 - You heard that right, we are giving your domains on Cloudflare an AI Index. 🤖🔎 Creators now get the tool to make their content discoverable by AI, and AI builders get access to better data. All with fair compensation. Sign up for private beta today! https://t.co/jt961wFQG8 - **andrewlamb1111** (2025-09-26): https://x.com/andrewlamb1111/status/1971522247329513979 - CloudFlare's Distributed R2 SQL engine's is a pretty good exemplar of how to build a serverless database to process petabytes in seconds using @ApacheDataFusio and @ApacheParquet https://t.co/8QSk6fOwuQ - **threepointone** (2025-09-26): https://x.com/threepointone/status/1971490324393648382 - shipping something super experimental and buggy today it’s the most exciting thing I’ve worked on all year - **geteviapp** (2025-09-26): https://x.com/geteviapp/status/1971417111944929718 - Exactly! I indeed incorrectly remembered, it failed with that eval error. Would be awesome to fix. I implemented Firestore clone via REST which does work on Cloudflare, but it is annoying to support things like that. The app is this btw: https://t.co/faIbM6m3C2 The failure is here: https://t.co/P0Vof1eYTG - **samgoodwin89** (2025-09-26): https://x.com/samgoodwin89/status/1971411891227664405 - Alchemy-effect's new CLI for approving and applying a plan is coming along nicely. So good to have planning back! https://t.co/x5TY2I03am - **rrnn** (2025-09-26): https://x.com/rrnn/status/1971380805529453036 - this is such a unique sink https://t.co/XBs1irqKBI - **kristianfreeman** (2025-09-25): https://x.com/kristianfreeman/status/1971361781328248996 - We use Astro to build and deploy Cloudflare’s docs. In the over half-decade I’ve been here, we have rewritten the docs in a lot of frameworks, Astro is easily the best. It is time for you to start Astromaxxing https://t.co/ax6Cecs9nZ - **thdxr** (2025-09-25): https://x.com/thdxr/status/1971360604700225665 - anyone that actually uses cloudflare understands that this is the best announcement of the week https://t.co/B8n8Ffa96J - **mojombo** (2025-09-25): https://x.com/mojombo/status/1971357781946138656 - RedwoodSDK up and to the right! If you like @Cloudflare and you like @reactjs, but especially if you like to SHIP PRODUCT, you should put your 👀 on this. https://t.co/iBLjuoYm2s - **JustJake** (2025-09-25): https://x.com/JustJake/status/1971336988419936435 - @irvinebroque Bingo It’s evident this saves the customer time, but it also saves the sales team time They can work on enablement of the underlying customer’s maximum success. In consumption models, this incentive alignment is pure It’s a winners philosophy, and always been ours as well - **thdxr** (2025-09-25): https://x.com/thdxr/status/1971332073505030518 - everyone doing these highly produced launch videos now we did a video walking around in my backyard talking at my phone and it cost nothing, got more views and people actually understood the message if you are copying other people it's already too late - **FardeemM** (2025-09-25): https://x.com/FardeemM/status/1971328559575978168 - Couldn't be more excited for an update. CF is so great, sadly held back by design. This is huge. Long NET https://t.co/telHr3uFcF - **MikeNomitch** (2025-09-25): https://x.com/MikeNomitch/status/1971323568299573755 - Cloudflare Containers limits just went way up - over 10X higher! Any paid user can run concurrent instances up to 400GiB, 100 vCPU, and 2 TB of ephemeral disk - That's enough for 1000 instances of our smallest type. https://t.co/Azh1Vif5ya - **ritakozlov** (2025-09-25): https://x.com/ritakozlov/status/1971323212102602889 - there are forever cursed defense-against-the-dark-arts positions and there are forever blessed pm-of-cloudflare-workers-runtime positions @irvinebroque @MikeNomitch_CF (and me) all started there, and have all had 📈 careers at cloudflare and we're hiring for it again 👇 - **MikeNomitch** (2025-09-25): https://x.com/MikeNomitch/status/1971320496454238668 - Love being part of a team that can go this deep technically and does the right things to keep Workers secure. Great read if you want to learn more about V8, memory protection keys, or memory corruption attacks. https://t.co/JaEono4CZG - **elidourado** (2025-09-25): https://x.com/elidourado/status/1971320222125789313 - Not the most important issue but I am baffled that we allow USPS to spam American citizens billions of times per year. https://t.co/zLjxUggkj5 https://t.co/1V0qzz09Ro - **kkblinder** (2025-09-25): https://x.com/kkblinder/status/1971319199919391067 - Can't relate 🤷‍♀️ https://t.co/T1k7NKCWo9 https://t.co/MpIPd5Gksj - **TheVRNerd** (2025-09-25): https://x.com/TheVRNerd/status/1971316145958633910 - .@CloudflareDev @Cloudflare IS AMAZING. They released httpServerHandler which now allows you to use other http frameworks like @UseExpressJS and even @hapijs from a cloudflare worker! Here is an example for @hapijs. https://t.co/ww5N41jkTZ - **jonas** (2025-09-25): https://x.com/jonas/status/1971312071515205740 - @mwylde @ArroyoSystems @ApacheDataFusio Can we connect this to workers analytics engine? I just want a single API to dump any json blobs into. And for some period of time I’ll use cheap and fast clickhouse to query but later on it all gets stored in R2 and I can do slow expensive queries using this new product? - **sama** (2025-09-25): https://x.com/sama/status/1971297661748953263 - Today we are launching my favorite feature of ChatGPT so far, called Pulse. It is initially available to Pro subscribers. Pulse works for you overnight, and keeps thinking about your interests, your connected data, your recent chats, and more. Every morning, you get a custom-generated set of stuff you might be interested in. It performs super well if you tell ChatGPT more about what's important to you. In regular chat, you could mention “I’d like to go visit Bora Bora someday” or “My kid is 6 months old and I’m interested in developmental milestones” and in the future you might get useful updates. Think of treating ChatGPT like a super-competent personal assistant: sometimes you ask for things you need in the moment, but if you share general preferences, it will do a good job for you proactively. This also points to what I believe is the future of ChatGPT: a shift from being all reactive to being significantly proactive, and extremely personalized. This is an early look, and right now only available to Pro subscribers. We will work hard to improve the quality over time and to find a way to bring it to Plus subscribers too. Huge congrats to @ChristinaHartW, @_samirism, and the team for building this. - **erikcorry** (2025-09-25): https://x.com/erikcorry/status/1971283459189719478 - When I'd written the first draft of my blog post on hardening security in Cloudflare Workers, one of the reviewers said it would be helpful to explain how a memory corruption attack on V8 worked pre-sandbox. https://t.co/haXqJhh3HZ - **dok2001** (2025-09-25): https://x.com/dok2001/status/1971280577153335383 - @ritakozlov @zenorocha The only competitor here is AWS SES. - **ritakozlov** (2025-09-25): https://x.com/ritakozlov/status/1971279644566618425 - email is a big market and as you said, not winner takes all. our goal is to offer all the building blocks developers need on our platform, but developers often choose to go multi provider to get best of breed (eg we have d1 but see planetscale announcement today) you guys have set precedence for what good dx for email looks like and we’re huge fans! - **0xPaulius** (2025-09-25): https://x.com/0xPaulius/status/1971276030376923217 - ok i did NOT expect cloudflare to ship so hard 🥶 https://t.co/knfmfn0tlF - **backpinelabs** (2025-09-25): https://x.com/backpinelabs/status/1971266089096184093 - This is going to be a really big deal for Cloudflare's dev platform and I am so excited about this launch! https://t.co/UVImrwnSoe - **aboodman** (2025-09-25): https://x.com/aboodman/status/1971264134470828403 - Still processing my trip back to SF. Haven't been there in three years. The transformation is remarkable. If you don't live in SF your mental model is probably out of date, it has rebounded incredibly strong. TBH the city is looking better and feels more vibrant than any of the 15 years I lived there from 2004-2019. But the most shocking/world-altering thing is the waymos. It is hard to overstate how big of a change to the feel of the city they are, or the way of life. People are spending too much time thinking about agents and not enough thinking about driverless cars, IMO. These things will change every aspect of daily life. It's as big a change as the iPhone. - **gingerhendrixai** (2025-09-25): https://x.com/gingerhendrixai/status/1971256891369279991 - ❤️I have been wanted stagehand support since forever. I even have a repo with a fork where I tried to hack it in myself. https://t.co/8otEK7F8GE - **rohinlohe** (2025-09-25): https://x.com/rohinlohe/status/1971255134958342222 - @irvinebroque lovely bucket names! - **danielbergholz** (2025-09-25): https://x.com/danielbergholz/status/1971253840436674985 - I feel like this is the closest we will ever get to convention over configuration in JS land 🔥 https://t.co/2Y6gEgXcyR https://t.co/TETQpKnOxy - **dan_jenkins** (2025-09-25): https://x.com/dan_jenkins/status/1971251912717836354 - @irvinebroque So pleased with this announcement - **jeromeschneider** (2025-09-25): https://x.com/jeromeschneider/status/1971248141694562765 - I worked on this! If you want to join the team, send me your CV! https://t.co/SPqNybAOkk - **dok2001** (2025-09-25): https://x.com/dok2001/status/1971246027186504166 - This has been a 7-year journey to making post-quantum cryptography safe to deploy at scale. Almost 30% of the top 100k sites are quantum safe because of @Cloudflare. Cloudflare's research team has been leading the push from standards to deployment. https://t.co/85XpoVXQoU https://t.co/IDR9j6rLA4 - **levelsio** (2025-09-25): https://x.com/levelsio/status/1971244547981070758 - Cloudflare today launched their own Email Sending service VERY cool, because by default every email service we use for sending for our websites gets acquired by private equity or IPOs and goes to shit It happened to MailChimp, then SendGrid, then Postmark It's just the nature of the beast, it's not a good business to be in, but it's essential for any website to be able to send emails @Cloudflare is so big ($75B) that for them it can just be a service as part of their general offering Because in 2025 maybe email sending should just be a feature, not a company P.S. please let me use Cloudflare Email with a basic JSON API, preferrably SendGrid/Postmark-compatible - **CloudflareTV** (2025-09-25): https://x.com/CloudflareTV/status/1971243252880683325 - Ready to dive deep into R2 SQL? Learn how Cloudflare's new distributed query engine works under the hood, from tackling I/O and compute challenges to using Apache Iceberg and DataFusion to deliver speedy, serverless queries. Watch now! https://t.co/29jSFqRotA https://t.co/h5HBYpCL8n - **MikeNomitch** (2025-09-25): https://x.com/MikeNomitch/status/1971241580414763128 - Huge progress on Workers compatibility with Node.js over the last year. - Run express.js or Koa.js - Read files from your bundle. Write tmp files. - DNS lookups (using 1.1.1.1 under the hood) - Way more packages "just work" - especially excited for SDKs that used node:http https://t.co/Rfs7GRiteH - **mwylde** (2025-09-25): https://x.com/mwylde/status/1971238635862376742 - Ok, this is absurd. We literally gave them the docs for R2 SQL yesterday. They got it working in marimo notebooks in less than an hour. https://t.co/ZTM0GBsu22 - **ketanhwr** (2025-09-25): https://x.com/ketanhwr/status/1971237667187912731 - It's been an absolute pleasure working on this, and learning from the legend @erikcorry :) https://t.co/DzUh8tTCv4 https://t.co/TjWEEqX2U4 - **marimo_io** (2025-09-25): https://x.com/marimo_io/status/1971234683599069236 - The @CloudflareDev Data Platform makes it easy to ingest data into an zero-egress data lake and query it at petabyte scale using serverless SQL — and marimo is the best way to interact with it. https://t.co/9YfvSofF9k https://t.co/7amzUe6FYF - **LakeAustinBlvd** (2025-09-25): https://x.com/LakeAustinBlvd/status/1971234642738176217 - except Cloudflare. it's Birthday Week and I just got a text message that read "hey be ready - a blog post draft is going to come in hot tonight for a new product we are releasing tomorrow that we built a few days ago" https://t.co/ntVVyqRlJC - **sidborderwala** (2025-09-25): https://x.com/sidborderwala/status/1971233147250761791 - Guys! Cloudflare is fixing their UI. https://t.co/O5Hp7d9PKd - **mwylde** (2025-09-25): https://x.com/mwylde/status/1971232148041674876 - We're also hiring across all of these teams! If you're a crack Rust engineer or have experience working on query engines or stream processing systems I'd love to chat. https://t.co/NqBvDhc6xl - **ritakozlov** (2025-09-25): https://x.com/ritakozlov/status/1971232024049614872 - let's have coffee and talk about day 4 of cloudflare birthday week ☕️, from home edition PS: going to do a bonus episode to cover technical deep dives (ERR_TOO_MUCH_CONTENT) https://t.co/cXTvbT8gXq https://t.co/t110qseYj5 - **eastdakota** (2025-09-25): https://x.com/eastdakota/status/1971226533831582073 - Introducing NET Dollar: https://t.co/D5QrMDx0qG. - **code_kartik** (2025-09-25): https://x.com/code_kartik/status/1971226221263593887 - Now you can make your own v0, bolt, loveable. Just use vibe sdk by cloudflare. https://t.co/TkoBMaxGl2 - **PlanetScale** (2025-09-25): https://x.com/PlanetScale/status/1971225785030803753 - We've partnered with @Cloudflare to make it easy to ship full-stack apps backed by Postgres or Vitess and Cloudflare Workers. Sign in to your PlanetScale or Cloudflare account to get started and scale to millions. https://t.co/NH7Jp7YJ0P - **eastdakota** (2025-09-25): https://x.com/eastdakota/status/1971225537839509537 - Getting back to our self-service roots (which, it turns out, enterprises also love)… https://t.co/9p6MnnOKsA - **thomasgauvin** (2025-09-25): https://x.com/thomasgauvin/status/1971225413356835204 - Kumo is the reason for the big improvements to Cloudflare dashboard. It makes it easy for ANY team to ship GREAT UX 🚀 Thank you @BraydenWilmoth & the @outerbase team & @nandafyi for bringing your relentless pursuit of better developer experiences https://t.co/gQLCNZcliC - **BraydenWilmoth** (2025-09-25): https://x.com/BraydenWilmoth/status/1971221265613816173 - After joining Cloudflare we introduced a new component library internal to the company – Kumo. This new Workers landing page uses that component library (and many other Developer Platform products). But most of the credit here goes to @nandafyi for literally COOKING 🧑‍🍳 https://t.co/4d8OSdxgtw - **KentonVarda** (2025-09-25): https://x.com/KentonVarda/status/1971220801132298497 - Super-wonky blog post from @erikcorry and co. about how we're continually hardening the Workers Runtime. Some deep systems engineering here. https://t.co/Y0lPjEwUci - **MarcSelwan** (2025-09-25): https://x.com/MarcSelwan/status/1971220574904123865 - 𝗥𝟮 𝗗𝗮𝘁𝗮 𝗖𝗮𝘁𝗮𝗹𝗼𝗴: Now includes managed compaction! This is especially timely with the Pipelines launch since streaming ingest tends to create a lot of small files and this will help keep up read performance. https://t.co/NoLmgB8PC5 - **MarcSelwan** (2025-09-25): https://x.com/MarcSelwan/status/1971219255426458058 - It's 3:30AM in Hawaii right now and I can't sleep because I'm too excited about this! The Cloudflare Data Platform! A collection of data infrastructure tools that let you ingest, store, and query analytical data right within Cloudflare that include 👇 https://t.co/ZCWw9Mbrnv - **TysonTrautmann** (2025-09-25): https://x.com/TysonTrautmann/status/1971218718274502663 - Cloudflare is the best place to park your data. Be sure to give this a try and LMK what you think! https://t.co/L8suirzCfd - **mscccc** (2025-09-25): https://x.com/mscccc/status/1971217915824542129 - My life is fun https://t.co/SJh5ao3BFC - **aninibread** (2025-09-25): https://x.com/aninibread/status/1971215965435331002 - I won't let this one get lost in the parade! 🎊 R2 infrequent access is now battle tested and in GA. If you have long-tail data that is accessed less frequently, you can put it in IA to save on storage costs. 💰 https://t.co/VrPUdLNdBp - **Cloudflare** (2025-09-25): https://x.com/Cloudflare/status/1971215681635910038 - Just shipped: we've partnered with PlanetScale to make deploying full-stack, database-backed apps using Workers + PlanetScale Postgres (or Vitess!) really, really easy. Link your accounts, create a config, and deploy an app that can scale to millions of QPS without having to think twice. https://t.co/1WSK5DmzuJ #BirthdayWeek - **whoiskatrin** (2025-09-25): https://x.com/whoiskatrin/status/1971213815216996488 - my first birthday week at @cloudflare and the energy is unreal. no challenge is too wild. we’re moving mountains right now to raise the bar on so many levels. grateful to be here and to work with such incredible people - **dok2001** (2025-09-25): https://x.com/dok2001/status/1971213646627020939 - New @Cloudflare workload alert! https://t.co/iRFcPA7pWZ - **kathyyliao** (2025-09-25): https://x.com/kathyyliao/status/1971213025496662458 - new from Browser Rendering for @CloudflareDev birthday week 🎂 1) Playwright support is GA 2) @Stagehanddev support in Beta 3) higher limits so you can scale more 👇 https://t.co/GkZ4syuUX6 https://t.co/VUnj4w4P4c - **celso** (2025-09-25): https://x.com/celso/status/1971210449774575790 - We announced Email Routing in 2021. It quickly became one of our most popular features. Today, we are announcing Email Sending, a new capability that allows you to send transactional emails directly from Cloudflare Workers. https://t.co/VxjP9VmG5p - **elithrar** (2025-09-25): https://x.com/elithrar/status/1971210379268399549 - surprise: you can now ingest, manage schemas and *query data* directly from within @cloudflare ⚡️ ingest JSON, query Parquet, transform it on the fly, and query it directly from your R2 bucket: https://t.co/rrYZFYH0vx https://t.co/uH71o6tqWr - **LambrosPetrou** (2025-09-25): https://x.com/LambrosPetrou/status/1971210287819956578 - This is huge! So many workloads and use-cases unlocked now for the Cloudflare Developer Platform. https://t.co/pwdPBUkKdj - **smallshen2** (2025-09-25): https://x.com/smallshen2/status/1971210264533139671 - @ritakozlov @thomasgauvin @Cloudflare Easier to build email ai agents on workers ! - **kristianfreeman** (2025-09-25): https://x.com/kristianfreeman/status/1971210244417257946 - We’re announcing a private beta for Cloudflare’s new Email Service today. In the past, you could process emails in a Worker, now you can send emails too. Crucial app infrastructure in the same easy-to-use bindings style developers love. Can’t wait to build with this. https://t.co/vvZ5hJ2Zzw - **Cloudflare** (2025-09-25): https://x.com/Cloudflare/status/1971209410715303986 - Over the year we have greatly expanded Node.js compatibility. There are hundreds of new Node.js APIs now available that make it easier to run existing Node.js code on our platform. https://t.co/wukxzMAYoc #BirthdayWeek - **mwylde** (2025-09-25): https://x.com/mwylde/status/1971209376531972543 - The news is finally out! Cloudflare has a Data Platform! We're starting with serverless streaming pipelines (powered by @ArroyoSystems), a managed Iceberg Catalog, and a new distributed SQL engine built on @ApacheDataFusio. https://t.co/nzjgS7Q9X8 - **mwylde** (2025-09-25): https://x.com/mwylde/status/1971209378952004024 - The result is the Cloudflare Data Platform, available today in open beta. See the blog post for the details and how to get started: https://t.co/bPlleMLxZN - **celso** (2025-09-25): https://x.com/celso/status/1971208921940623545 - You can now run Stagehand with Browser Rendering and Workers AI https://t.co/2VmGzAwH0W - **elithrar** (2025-09-25): https://x.com/elithrar/status/1971207514629996740 - couldn't be more excited for this one. private beta coming in November. just a few more things to wrap up! a few lines of code to send an email from your Worker, with observability built right in: https://t.co/wFnco2NK5h https://t.co/peIArlyDAD - **CloudflareDev** (2025-09-25): https://x.com/CloudflareDev/status/1971205738908484057 - 📣 This is not a drill: Email Sending is coming to Workers: just call email.send() - ! As part of this, we're combining Email Sending and Email Routing together into a cohesive, unified hub to manage emails. - **threepointone** (2025-09-25): https://x.com/threepointone/status/1971205385987219677 - .@cloudflare ⨉ @PlanetScale probably nothing https://t.co/At8wgiP7H2 https://t.co/UTDS2FD9ac - **thomasgauvin** (2025-09-25): https://x.com/thomasgauvin/status/1971205069204074712 - emails are coming to Cloudflare! excited to be getting this into your hands asap to make it easier to rely on email as a core layer of your application (that's how we view emails! read the blog) https://t.co/9Jvpo2jPJz - **CloudflareDev** (2025-09-25): https://x.com/CloudflareDev/status/1971204725581414436 - Say hi to the Cloudflare Data Platform — a serverless data platform built on Cloudflare's global infrastructure. Ingest data with Pipelines, manage it with our Iceberg-compatible Data Catalog, and *query it* using the brand-new R2 SQL query engine. - **michellechen** (2025-09-25): https://x.com/michellechen/status/1971203385333563534 - ai should be accessible to all, and that means meeting people where they are introducing some new models developed by local ai labs in India Japan, and South East Asia region. excited to support new languages and capabilities on Workers AI https://t.co/EKzN7nDHKH - **CloudflareDev** (2025-09-25): https://x.com/CloudflareDev/status/1971203162015973802 - 📢Announcing R2 SQL in open beta! We launched our in-house distributed SQL query engine for efficient queries over massive amounts of data in R2 Data Catalog. It's designed to be completely serverless and take advantage of Cloudflare’s distributed compute and global network. - **CloudflareDev** (2025-09-25): https://x.com/CloudflareDev/status/1971202336325619973 - 📣Big news: Cloudflare Pipelines is now in open beta. You can now use simple SQL queries to process and transform real-time data (logs, clickstream etc) directly to R2 Data Catalog! We’re excited for you to try the beta and share your feedback! - **southpolesteve** (2025-09-25): https://x.com/southpolesteve/status/1971201792903086098 - You'll have to wait a bit longer to see Cloudflare Ornithologist. Apologies to all the bird lovers out there. I promise its gonna blow your mind. https://t.co/dgFnln4XOi - **samlambert** (2025-09-25): https://x.com/samlambert/status/1971201730374402366 - "We’ve teamed up with PlanetScale because we wanted to partner with a database provider that we could confidently recommend to our users: one that shares our obsession with performance, reliability and developer experience. These are all critical factors for any development team building a serious application." https://t.co/qkbbQqy6AB - **threepointone** (2025-09-25): https://x.com/threepointone/status/1971201221395665146 - self-serve for server-less https://t.co/KCfYPmOamW - **elithrar** (2025-09-25): https://x.com/elithrar/status/1971201189296603409 - just shipped: link your Cloudflare & @PlanetScale accounts + connect Workers to your Postgres or Vitess databases in a couple of clicks. https://t.co/BOfhA5umoY https://t.co/Z6RABVLdnf - **ritakozlov** (2025-09-25): https://x.com/ritakozlov/status/1971200868184740135 - i am in tweet paralysis because sooo many things went out today but big one: transactional email is coming to a @cloudflare worker near you 📧 https://t.co/HEMFm8w50H - **aninibread** (2025-09-25): https://x.com/aninibread/status/1971200596276629845 - Notice anything different on the dashboard? 👀 AI Search == AutoRAG, now letting you bring your own provider keys and get access to even more models to use for inference and embedding. 🪄 https://t.co/BQ4L6tesWD - **yagiznizipli** (2025-09-25): https://x.com/yagiznizipli/status/1971200183867494777 - you asked, we delivered :-) https://t.co/aDfHoLAUuc - **erikcorry** (2025-09-25): https://x.com/erikcorry/status/1971200164661743690 - https://t.co/YECEAAQD71 - **erikcorry** (2025-09-25): https://x.com/erikcorry/status/1971200109263438293 - We have been hardening Cloudflare Workers against attacks. https://t.co/JfKrc1zJPo - **Cloudflare** (2025-09-25): https://x.com/Cloudflare/status/1971199273422258678 - Over the next year Cloudflare will make nearly every feature we offer available to any customer who wants to buy and use it. https://t.co/YAt2DoO8Du #BirthdayWeek - **appfactory** (2025-09-25): https://x.com/appfactory/status/1971123476607140218 - Two days ago it was 6,000 weekly installs! https://t.co/7czkDmngkE - **eastdakota** (2025-09-25): https://x.com/eastdakota/status/1971098649909395927 - This is such a no-brainer feature in retrospect. The one upside of the Salesloft breach was it made it blindingly obvious we needed to build this. https://t.co/3w7CM1Fr7r - **waleedalrashedd** (2025-09-25): https://x.com/waleedalrashedd/status/1971069381183668458 - Loving the new polish on the @Cloudflare 'Workers & Pages' dashboard. The consolidated view is a smart move, and the entire UI feels much more responsive. Great to see the focus on refining the developer workflow. thanks @CloudflareDev 👏 #Serverless #WebDev https://t.co/MLYS5dpC6a - **thdxr** (2025-09-25): https://x.com/thdxr/status/1971043122575806552 - this is why you think open source models aren't good on the left is the average provider, on the right is opencode zen left is slow, vomits out tool call parse failures, confuses itself and takes 5x longer to complete https://t.co/9Elf5GkTWC - **natolambert** (2025-09-25): https://x.com/natolambert/status/1971028376849920313 - This is a very cool release. At the same time I'd like to see Meta be less cautious in their legal strategy around the release. The boundaries between the research & product communities are smaller than they've ever been, and noncommercial licenses really hamper innovation on top of research. This would really not hurt anything Meta is doing to be released with a permissive license, just seems like lawyers have the final say. Novel architectures alone slow uptake a lot! It takes a long time to get these changes adopted. Don't put up walls to research like this. - **nichochar** (2025-09-25): https://x.com/nichochar/status/1971010315061195143 - thats a buy https://t.co/TmM6Ik2MUk - **kentcdodds** (2025-09-24): https://x.com/kentcdodds/status/1970996692422922705 - This is super cool: https://t.co/5FskG4kzVY Now your editor/agent can debug your web app directly using Chrome DevTools 🔥 https://t.co/OKEreFP2DP - **yagiznizipli** (2025-09-24): https://x.com/yagiznizipli/status/1970988477732012307 - I am excited about tomorrow’s @cloudflare birthday week announcement! - **garrytan** (2025-09-24): https://x.com/garrytan/status/1970963654817480738 - They'll do anything to block housing. ANYTHING. https://t.co/dm4GvnyBjX - **AIatMeta** (2025-09-24): https://x.com/AIatMeta/status/1970963571753222319 - New from Meta FAIR: Code World Model (CWM), a 32B-parameter research model designed to explore how world models can transform code generation and reasoning about code. We believe in advancing research in world modeling and are sharing CWM under a research license to help empower the community to build upon our work. ➡️ Read the technical report: https://t.co/i9BqtfyJ7L ➡️Download the open weights: https://t.co/S2CxqCOMn0 ➡️Download the code: https://t.co/wOsDR8Q5OQ - **syhw** (2025-09-24): https://x.com/syhw/status/1970960847494316168 - 5/ The team and I can’t wait to see what new research will be enabled with a world model. We release 3 checkpoints under a research license: 1️⃣ CWM pretrained, e.g. for new posttrainings methods, 2️⃣ CWM SFT, e.g. for RL research, 3️⃣ CWM, e.g. for inference time scaling. https://t.co/jcmBuXDin2 - **syhw** (2025-09-24): https://x.com/syhw/status/1970960842586980505 - 3/ CWM allows us to study this question. Our model is trained on large amounts of coding data & bespoke Python + Bash world modeling data, allowing it to simulate Python function execution and agentic interactions in Bash environments. - **syhw** (2025-09-24): https://x.com/syhw/status/1970960844449251503 - 4/ Here is an example of the Code World Model tracing the execution of the piece of code counting the "r"s in "strawberry". Think of it like a neural `pdb` that you can set to any initial frame state, and that reasoning can query as a tool in token space. https://t.co/hVb79hKJSG - **syhw** (2025-09-24): https://x.com/syhw/status/1970960837721653409 - (🧵) Today, we release Meta Code World Model (CWM), a 32-billion-parameter dense LLM that enables novel research on improving code generation through agentic reasoning and planning with world models. https://t.co/BJSUCh2vtg - **syhw** (2025-09-24): https://x.com/syhw/status/1970960839596441999 - 2/ When humans plan, we imagine the possible outcomes of different actions. When we reason about code we simulate part of its execution in our head. The current generation of LLMs struggles to do this. What kind of research will an explicitly trained code world model enable? https://t.co/0dnT7ph8Ri - **FredKSchott** (2025-09-24): https://x.com/FredKSchott/status/1970934724215570504 - It will always feel weird watching the $$$ increase as I use opencode, claude code, etc. I can't think of any other piece of software that works like this. like watching the pump as you fill your car with gas - **ritakozlov** (2025-09-24): https://x.com/ritakozlov/status/1970873894879047982 - let's have coffee and talk about day 3 of cloudflare birthday week ☕️ https://t.co/hM4nv1T9QN https://t.co/ZJgZbPOgrO - **tomosman** (2025-09-24): https://x.com/tomosman/status/1970873659138187683 - Build your own "vibe coding" platform using VibeSDK from Cloudflare https://t.co/UsWDQ7ea3S - **ccheever** (2025-09-24): https://x.com/ccheever/status/1970861337996836910 - Liquid glass, link previews, context menus, and more. Expo Router v6 gives you more native features that wow than any other navigation system. https://t.co/LvrWZ3gdBY - **DKThomp** (2025-09-24): https://x.com/DKThomp/status/1970840029254815970 - First ever successful treatment of Huntington’s disease! LFG https://t.co/4WPoLUkXBD https://t.co/1OjKAfQ3gU - **nickgrossman** (2025-09-24): https://x.com/nickgrossman/status/1970839380173668618 - I spoke w/ @cloudflare co-founder & CEO Matthew Prince (@eastdakota) for the latest ep of the Slow Hunch podcast. Since 2010, Matthew and his team have built Cloudflare into one of the most important companies on the internet: powering and protecting vast portions of global traffic. I’ve known Matthew since @USV’s investment in Cloudflare’s Series C back in 2013, and it was really fun tracing his slow hunch, right from those early days. We covered: - how he met co-founders Michelle Zatlyn (@zatlyn) and Lee Holloway (@icqheretic) - the early experiments and risks that shaped Cloudflare - stories from the company’s pre-IPO days - their decision to make website encryption free - how their infra ended up running 2 of the internet’s 13 root servers - his thoughts on AI (esp the transition from a search-driven internet to an answer-driven one) Hope you enjoy! - **Zephyr_hg** (2025-09-24): https://x.com/Zephyr_hg/status/1970812569771196750 - 5/ Cloudflare open-sourced VibeSDK for one-click AI coding platforms. Anyone can deploy AI-powered coding platforms instantly. Build websites by describing what you want in sentences. Includes isolated sandbox environments and global deployment. https://t.co/z3Zh3Uv5mW - **liadyosef** (2025-09-24): https://x.com/liadyosef/status/1970810786747732428 - Wow 🤯 Mini life goal achieved - having the the amazing @kentcdodds recording an *entire series* about MCP-UI. This is part of his awesome MCP workshop, be sure to check it out! As always with Kent's workshops - this is the best way to becoming an MCP wizard. (+ @idosal1) https://t.co/9zTA1yhS0l https://t.co/OwUxuLvJet - **collision** (2025-09-24): https://x.com/collision/status/1970783012813160569 - I always enjoy @destraynor’s no-nonsense insights. He and I discuss growing @Fin_ai to $50 million revenue run rate, his “four horseman” rubric for investing in AI companies, advice for product marketers, and why you shouldn’t split the G. Subscribe on Spotify, Apple Podcasts, and YouTube for future Cheeky Pint episodes. Timestamps 00:00 Intro 02:58 Reinventing Intercom 06:31 Fin 18:06 1M resolutions a week 24:22 Selling AI 29:34 Product marketing 37:18 Listening to users 44:14 Usage-based billing 45:14 Advice for startups 52:09 AI pricing 01:07:27 Co-founder dynamics 01:11:04 Predictors of company success 01:15:56 How AI-native is Intercom? - **thesquashSH** (2025-09-24): https://x.com/thesquashSH/status/1970660354436014579 - @irvinebroque (not to diminish the amount of effort it takes to build this stuff, it's hard to get right, I just mean it's an old design pattern) - **thesquashSH** (2025-09-24): https://x.com/thesquashSH/status/1970660133589156042 - @irvinebroque basic ORM shit lol, Django and other remote resource adapter libs have been doing stuff like this for decades to reduce n+1 - **dok2001** (2025-09-24): https://x.com/dok2001/status/1970657890894782552 - Lucky to get to work with @threepointone - **bshlgrs** (2025-09-24): https://x.com/bshlgrs/status/1970657863749247199 - @austinc3301 @irvinebroque You might enjoy learning generally about monads if you like this :) - **DhravyaShah** (2025-09-23): https://x.com/DhravyaShah/status/1970608691155214726 - Github MCP has 80 tools Literally why would anyone use it 😭 mcp devs should be thoughtful about any and all tools added - **southpolesteve** (2025-09-23): https://x.com/southpolesteve/status/1970592567088562293 - I'm hiring an eng manager for our runtime languages/APIs. Super cool role if you want to work with folks deep in JS, Node, Rust, WASM, Python. You’ll be helping define how devs actually use Workers. Some of the most impactful APIs you’ll ever touch. https://t.co/XR2ZmEpPA8 - **collision** (2025-09-23): https://x.com/collision/status/1970586725622915144 - Coming up tomorrow on Cheeky Pint: @destraynor from @intercom 🇮🇪 Two pints' worth of essential advice for every product leader: building great software, the challenges of pricing AI products, and having empathy for procurement people. https://t.co/1nnYJdOMnM - **tannerlinsley** (2025-09-23): https://x.com/tannerlinsley/status/1970579559260262587 - 🚀 Announcing @Tan_Stack Start v1 Release Candidate! Upgrades ↓ ✨ Unified Route Tree: no more server-specific files 🔐 Type-safe middleware & server context upgrades 🛡 CSP/nonce support ⚡ Now works with any native Vite Env plugin 🌀 Zero-JS: any server handler can render! - **dok2001** (2025-09-23): https://x.com/dok2001/status/1970579143478710516 - Better @Cloudflare integration https://t.co/XExpMjyvqI - **roerohan** (2025-09-23): https://x.com/roerohan/status/1970578490652270918 - If you're looking to build ambitious products with our kickass team, DM me or reach out to @KankaniAbhishek https://t.co/aWZEEIZQ9f - **ImSh4yy** (2025-09-23): https://x.com/ImSh4yy/status/1970551650550194209 - This is quite a big deal. These companies jointly funding Astro and TanStack and collaborating to keep these frameworks platform-agnostic. No single company controlling the direction. No vendor lock-in. Proper open source. https://t.co/5xcU4H1xZQ - **FredKSchott** (2025-09-23): https://x.com/FredKSchott/status/1970543144224858408 - Cloudflare's blog post on this nails it IMO and is a huge validation on the original idea behind Astro: Client-side SPAs are great for dynamic, complex web apps. Tanstack is a joy to use. It pairs great with Astro when you need dynamic content (dashboards, search pages, web apps, etc). I genuinely love it. Client-side SPAs fall apart for high-performant content sites. Too much JavaScript, poor performance, poor SEO, etc. Content fails when you treat it like a big application! Older JavaScript frameworks tried to solve this by throwing complexity at the problem. SSR, ISR, PPR, client/server components, Isomorphic rendering... but complexity comes at a cost, and we're still here waiting for some of these promised solutions/bets to pay off. The reality is that client-side SPAs are actually great! Simplicity is a strength! A great web app built with Tanstack Router + local-first Tanstack DB will feel faster than any server-rendered application. The performance and complexity cost of adding SSR to a web app isn't always worth it. Astro and Tanstack are two sides of the same coin. We have our strengths and weaknesses, but together both cover the entire spectrum of what you might want to build on the web. - **leinwand** (2025-09-23): https://x.com/leinwand/status/1970532797552173258 - 🚢🚢🚢 Why @Cloudflare, @Netlify, and @Webflow are collaborating to support Open Source tools like Astro and TanStack by @ritakozlov https://t.co/GbuOJZ4MsQ - **biilmann** (2025-09-23): https://x.com/biilmann/status/1970517954820211008 - At Netlify we’ve taken on the “Preferred deployment partner” sponsorships for both @astrodotbuild and @tan_stack over the last years, but today we’re genuinely excited to have @Cloudflare join us and drop any notion of preferred labels. Open source thrive when we can all invest together across the industry and across competitive barriers, and these projects will lift the whole open web if well funded. Hope more partners will join in and help grow the open web ecosystem. - **ritakozlov** (2025-09-23): https://x.com/ritakozlov/status/1970509656964424150 - let's have coffee and talk about day 2 of cloudflare birthday week together ☕️ it's a long boi bc i talk about x402 and bots for a while 🤖 https://t.co/3pfc8wfJt8 https://t.co/NlYILu2mod - **ChromiumDev** (2025-09-23): https://x.com/ChromiumDev/status/1970505063064825994 - Announcing our public preview of Chrome DevTools MCP! Experience the full power of DevTools in your AI coding agent → https://t.co/u5kOrjFgrq With Chrome DevTools MCP, your AI agent can run performance traces, inspect the DOM, & perform real-time debugging of your web pages. https://t.co/aKgmArbihX - **Cloudflare** (2025-09-23): https://x.com/Cloudflare/status/1970503597268894084 - Cloudflare is doing anomaly detection like never before. Read about how we build and deploy bespoke, per-customer models to detect traffic anomalies and catch more sophisticated bots in bot management and Super Bot Fight Mode. https://t.co/smpyyhDWmH #BirthdayWeek - **eastdakota** (2025-09-23): https://x.com/eastdakota/status/1970503464120697311 - Super important that core protocols have independent governance. Excited to partner with @Coinbase to bring independent governance to the new x402 protocol. https://t.co/bmbE04NCS5 - **dok2001** (2025-09-23): https://x.com/dok2001/status/1970500634387636629 - A second chance for 402 status code to make its mark on the Internet. The original creators understood that payments needed to be built into the network infrastructure. https://t.co/LS6gX4tCcD - **rohinlohe** (2025-09-23): https://x.com/rohinlohe/status/1970499757295780179 - We @Cloudflare are partnering with @Coinbase to create the x402 Foundation to build the future of Internet-native payments and drive adoption of x402. https://t.co/KNBxGciaCE - **DrizzleORM** (2025-09-23): https://x.com/DrizzleORM/status/1970499038660133247 - 👀👀👀👀👀👀👀👀👀👀👀👀👀👀👀👀👀👀👀👀👀👀👀👀👀👀👀👀👀👀👀👀👀👀👀👀👀👀👀👀👀👀👀👀 https://t.co/KPjANel7CS - **ritakozlov** (2025-09-23): https://x.com/ritakozlov/status/1970498608257536499 - cloudflare docs are powered by @astrodotbuild and we use many @tan_stack libraries like query in production excited for @tan_stack start 1.0 RC today excited about our partners in crime: @netlify 🧡 @tan_stack and @webflow 🧡 @astrodotbuild excited for the web 🧡🧡🧡 https://t.co/Lz8l9bYf2T - **tan_stack** (2025-09-23): https://x.com/tan_stack/status/1970495735348240596 - 🎉 @Cloudflare and @Netlify have been so great to work with. They care so much about the web and we're honored that they consider us a critical part of their vision. With their support, our Q4 is going to be 🔥 https://t.co/Xh7ekvszqe - **thomasgauvin** (2025-09-23): https://x.com/thomasgauvin/status/1970495690704069078 - This is actually incredibly impressive, hats off to the team 🔥🚀 An end-to-end vibe coding application, available *as a template* for you to get started building your own vibe coding apps (!!!) Plus, a perfect example of sandboxes, ai gateway and workers for platforms in one https://t.co/0fe3P5IHav https://t.co/7i2JNNnMlG - **burcs** (2025-09-23): https://x.com/burcs/status/1970493062188278064 - i've been dreaming of a cloudflare vibe coding platform forever, and not only did we ship one, we open-sourced it so you can run it yourself! or better yet host it and build your own vibe coding company... 🤩 https://t.co/Qc5WLbP09n - **threepointone** (2025-09-23): https://x.com/threepointone/status/1970492347281457245 - Cloudflare India _killing_ it in the design game rn //cc @Paramssharma @akigugale https://t.co/BAZxT6eKYO - **KankaniAbhishek** (2025-09-23): https://x.com/KankaniAbhishek/status/1970491662389727401 - This has end-to-end been built by the ETI team in India! Checkout the live demo at https://t.co/aYxRIWJAJT Keep an eye out for more exciting stuff that we’re about to announce :) Aaandd, we’re hiring! More details about the process to follow https://t.co/Yu3UUlVKEW - **yusukebe** (2025-09-23): https://x.com/yusukebe/status/1970490527499145439 - Cloudflareのx402のデモおもしれーー * Coinbaseと協力し「x402」を設立 * 402 Payment Requiredを使ってHTTP経由で支払いする * デモではMCPサーバーがx402を実装し支払いを要求 * クライアントも実装してある * 0.01ドルだけ使える * ツール呼び出し使う * 一度使うと無一文になるので使えない https://t.co/XK3iKOp9s2 - **elithrar** (2025-09-23): https://x.com/elithrar/status/1970489199540228156 - possibly my favorite announcement (it's only tuesday?!) deploy your own multi-user vibe-coding platform — for your internal teams, as the basis for something new, or to bring code gen to an existing app — and run it on @cloudflare Workers. https://t.co/iQn8rhu2vs https://t.co/K7LZ4gGpsq - **threepointone** (2025-09-23): https://x.com/threepointone/status/1970487870084522041 - was fretting that we don't have a codegen tool, so they made a whole platform in a box for building codegen tools lmao https://t.co/OAhAjxDIQY - **CloudflareDev** (2025-09-23): https://x.com/CloudflareDev/status/1970485944542257494 - With support from our friends at @Netlify & @Webflow, we’re proud to back two essential open-source projects: @Astrodotbuild (for creating high-performance content sites) and @tan_stack (for building dynamic, complex web apps). We believe building the future of the web, together, and can’t wait to see what you build! - **kristianfreeman** (2025-09-23): https://x.com/kristianfreeman/status/1970485087100903434 - You can now build and deploy your own vibe-coding platform on top of @CloudflareDev. One-click deploy to get started: https://t.co/BugWYwe4US - **williamallen** (2025-09-23): https://x.com/williamallen/status/1970484728915730891 - We @Cloudflare are working with @coinbase to create the x402 Foundation. @programmer & others at Coinbase have created a powerful protocol to enable payments at Internet scale, and I am proud to work with them to create a foundation to standardize the protocol. Every day, sites on Cloudflare send out over a billion HTTP 402 response codes to bots, crawlers & agents. But without a standard, open way to respond to these 402s, they too often go unanswered. x402 can help solve this. (link in next tweet...) - **TarushNagpal** (2025-09-23): https://x.com/TarushNagpal/status/1970484205764387317 - Had a great time building https://t.co/UDwu6MyaPC 😬 Go check it out! https://t.co/YkySHg5MdM - **dinasaur_404** (2025-09-23): https://x.com/dinasaur_404/status/1970482339282329700 - You can now deploy a whole AI vibe coding platform in one click! The whole platform? Correct 🙂 We stitched together all the pieces, so you can take them apart and build your own AI vibe coding platform — for internal use or to build into your product. - **astrodotbuild** (2025-09-23): https://x.com/astrodotbuild/status/1970481062083187088 - Cloudflare just donated $150,000 to support Astro's open source development as our newest official hosting partner. Thank you @cloudflare! 🚀 - **whoiskatrin** (2025-09-23): https://x.com/whoiskatrin/status/1970477600331190721 - 🔥 major update: agents built using Agents SDK can now pay for resources with x402, and MCP servers can expose tools to be paid for via x402 stellar work by @caise_p and the whole team as always! you can try immediately: https://t.co/gVtBqs8x4s https://t.co/iTC2IsRtKk - **ritakozlov** (2025-09-23): https://x.com/ritakozlov/status/1970477257798951047 - the @cloudflare agents sdk now supports x402 which means you can build MCP servers and agents with built-in: → state / memory → auth → and importantly, payments (!!!) blog & docs 👇 - **CoinbaseDev** (2025-09-23): https://x.com/CoinbaseDev/status/1970477072792592484 - 🚀 Big news 🚀 @coinbase and @cloudflare will be launching the x402 Foundation to establish x402 as the universal standard for AI-driven payments. Legacy payment systems weren’t built for machines, but x402 changes that by embedding payments directly into web interactions, enabling AI agents, APIs, and apps to transact value as easily as they exchange data. This is a generational moment for payments. With the x402 Foundation, we’ll help build the infrastructure for an open economy where agents and people can transact seamlessly. - **Cloudflare** (2025-09-23): https://x.com/Cloudflare/status/1970474713294962823 - Today, Cloudflare is proud to announce support for two cornerstone frameworks in the modern web ecosystem: we’re partnering with Webflow to sponsor Astro, and with Netlify to sponsor TanStack. https://t.co/OWTQWK4kMU #BirthdayWeek - **robin_j_brooks** (2025-09-23): https://x.com/robin_j_brooks/status/1970456761153790322 - The US is gradually losing its "exorbitant privilege," its ability to issue debt more cheaply than the rest of the G10 when foreign yields are swapped back into $. That loss is most pronounced for longer-term yields where risk premia are building. US deficits are taking a toll... https://t.co/UWlFiUpnWr - **hite_sh** (2025-09-23): https://x.com/hite_sh/status/1970451732875616361 - Thrilled to be a part of Cloudflare Launchpad! https://t.co/WHxloaL6Gf - **appfactory** (2025-09-23): https://x.com/appfactory/status/1970413992683118944 - RedwoodSDK already have realtime and live pages via Durable Objects, but with Cap 'n Web you can have live components. @RedwoodJS are going to build "useState" via "useLiveState" to the server. It'll be bi-directional. So you can discretely update on a per-component level rather than the entire page. - **kamikaz1_k** (2025-09-23): https://x.com/kamikaz1_k/status/1970338279774937116 - pipelining RPCs in a native JavaScript API so you can DIY GraphQL resolver…and more? VERY cool What if you paired this with Effect or something for a very dynamic job builder? This is actually pretty interesting for agent workflows… https://t.co/UWEl32JbRQ - **madiator** (2025-09-23): https://x.com/madiator/status/1970300525381054555 - Let me try to explain why I don't really like RAG. But before I start, I want to say, RAG is a pretty useful tool and can address 95+% of the use cases, especially now that contexts are quite long and with newer techniques like agentic RAG. So the following text is mostly for understanding sake and to develop a different perspective. ## Point 1: Compositional generalization So, during pretraining, LLMs see all sorts of text. They learn about Shakespeare and they learn about programming, independently. And then they are able to write a python program with comments the style of Shakespeare, though they have not seen these two together. This is compositional generalization. LLMs are able to learn many different skills/facts during pretraining and then combine them together during inference, and this is not something that can happen easily with RAG especially due to the retrieval part. One ~visual analogy for thinking about RAG is as follows: you have a question about a book, and you ask a retrieval model to tear a page or two, which you use to read and answer. Tough luck if the answer is not there, or if the answer is spread across many many pages (e.g. "how did the character evolve over the different chapters"). ## Point 2: Compression <-> Understanding Feynman had famously said "What I can't create I don't understand". That's why we invented "generative" models, which compress all the knowledge in the world during pretraining and can then generate (after having understood the world). Search stands in the way. Long context is better, but it has it's own challenges. Agentic RAG and all the other techniques are mostly hacks to improve upon the situation. Techniques like building a knowledge graph etc., are very much heuristics and don't really generally work well across the spectrum. What has worked well for all of the Internet is to pretrain and then do SFT. So I hope we can do something similar with a small corpus, and that's what Mr. mcconaughey wants. - **rough__sea** (2025-09-23): https://x.com/rough__sea/status/1970291121256571268 - This looks so nice! I’m a massive fan of everything Kenton builds https://t.co/WEPDGD6rbq - **KentonVarda** (2025-09-22): https://x.com/KentonVarda/status/1970274215833002052 - I think adding the .map() thing was worth it just for the engagement on these threads lol. - **kellabyte** (2025-09-22): https://x.com/kellabyte/status/1970244628327879029 - “Merits of RPC have been subject to a great deal of debate. RPC is often accused of committing many of the fallacies of distributed computing. This reputation is outdated. When RPC was first invented some 40 years ago, async programming barely existed. “ https://t.co/HokUysLWmR - **grinich** (2025-09-22): https://x.com/grinich/status/1970231645098909980 - > Based on code introduced in today’s betas, we can confirm that Apple is laying the groundwork to bring MCP support to App Intents. 🤯 https://t.co/4OsqM9iW6D - **kkblinder** (2025-09-22): https://x.com/kkblinder/status/1970199642798526895 - no🧡 https://t.co/0qnDfgxn9h https://t.co/OUa4Ni6T32 - **zeeg** (2025-09-22): https://x.com/zeeg/status/1970198950830645408 - "immigrants are taking the jobs intended for me!" "also i refuse to relocate, think I'm hot shit, and have no desirable skills" - **jsensarma** (2025-09-22): https://x.com/jsensarma/status/1970194967802032510 - @arpitrage lot like China buying US Treasuries. - **southpolesteve** (2025-09-22): https://x.com/southpolesteve/status/1970150343737520405 - When I joined Cloudflare I barely understood Cap’n Proto. As I learned more I realized promise pipelining is basically what I always wanted out of something like GraphQL. Now we finally get to bring that power directly to the web. https://t.co/5AgX56LcIr - **Cloudflare** (2025-09-22): https://x.com/Cloudflare/status/1970133518953889980 - Cap'n Web is a new open source, JavaScript-native RPC protocol for use in browsers and web servers. It provides the expressive power of Cap'n Proto, but with no schemas and no boilerplate. https://t.co/DgT1Gyou4a #BirthdayWeek - **dok2001** (2025-09-22): https://x.com/dok2001/status/1970121565141372931 - I love big, audacious projects. @awesomekling’s @ladybirdbrowser, and @dhh's @OmarchyLinux are examples of entrenched areas that need diversity and new approaches. https://t.co/loxkkdl25m - **eastdakota** (2025-09-22): https://x.com/eastdakota/status/1970120851917099477 - While others are cutting their intern programs we’re doubling down on ours. We want to train the next generation of technical leaders and also take advantage of 1,111 fresh eyes as we go through the massive shift to an AI-enabled workplace. #BirthdayWeek https://t.co/YXZnMp5KDo - **CloudflareDev** (2025-09-22): https://x.com/CloudflareDev/status/1970117335198404919 - A better internet depends on a diverse ecosystem. Today, we're proud to announce our support for @ladybirdbrowser (a completely independent browser built from scratch) and @OmarchyLinux (making Linux more approachable for devs). Super excited to see what the community builds! - **korinne_dev** (2025-09-22): https://x.com/korinne_dev/status/1970117003818991715 - I previously worked on GraphQL tools, so I've been really excited about this Birthday Week blog! With Cap'n Web, you get the benefits of GraphQL (single network roundtrip for all your data, no overfetching) -- without needing to standup a separate server and schema. It's all typescript! - **threepointone** (2025-09-22): https://x.com/threepointone/status/1970113345698103362 - mindfuck library - define an api on your worker - call it from your frontend - handles Dates, Maps, Sets, whatever - handles functions and arrays (!) - automatic batching - inbuilt auth yes sir cap'n web🫡 https://t.co/VWfGIOGBRA - **KentonVarda** (2025-09-22): https://x.com/KentonVarda/status/1970112018553438394 - 2. In Cap'n Web, if you have a promise for an array, you can call .map() on it to apply a transformation to every element of the array. The way this works is WILD; see the blog post for more. https://t.co/oInyK7neuV - **KentonVarda** (2025-09-22): https://x.com/KentonVarda/status/1970112002803880239 - I don't have a problem with making new RPC systems! I can stop any time I want! But I don't want to. So I made a new one for you. It's called Cap'n Web. It's like Cap'n Proto RPC, except focused on working great in browsers and JavaScript/TypeScript servers. (Links in thread.) https://t.co/LAhZzXYUAM - **KentonVarda** (2025-09-22): https://x.com/KentonVarda/status/1970112005530128536 - Here's the blog post. Unlike Cap'n Proto, Cap'n Web does not use schemas. In fact, it requires almost no boilerplate whatsoever. You export a JS/TS API, you call it from the client. Just like Workers RPC, but in the browser, and with a few new twists. https://t.co/p0RBCT34FL - **thdxr** (2025-09-22): https://x.com/thdxr/status/1969955343246791043 - to be good at devtools you need to be a skilled programmer and have gone deep in a bunch of areas but you also cannot be too proud of this and derive your identity from it otherwise you will never make something that gets used by everyone - **dhaber** (2025-09-21): https://x.com/dhaber/status/1969805004363206859 - I’ve always believed that opportunities live between fields of expertise. If something fits neatly in a bucket, it’s probably already priced in. This has been my investment philosophy, a metaphor for my career and where I like to invest. Am I Silicon Valley or Wall Street, a founder, operator or investor? My answer has always been the same: Yes. - **robin_j_brooks** (2025-09-21): https://x.com/robin_j_brooks/status/1969780242970821063 - There's instability building in EM. Best way to see that is to look at how currencies have done since Powell's very dovish Jackson Hole speech on August 22. That speech should have lifted all EM. Instead, we're seeing many EM currencies struggling. Counterintuitive and worrying. https://t.co/PAcfMdzJ2r - **calebwatney** (2025-09-21): https://x.com/calebwatney/status/1969602836766900297 - Quick H-1B takes: 1) Current effort is (probably) not legally viable without Congress. 2) There's a WORLD of difference between applying a fee to applicants who are stuck in the lottery (capped employers) and to those who aren't (research nonprofits and universities). If we apply this rule to the latter, we will dramatically reduce the number of international scientists working in the U.S. 3) In theory, a fee just on *capped* employers can help differentiate between high-value and low-value applications, but 100k seems too high (especially if it's per year). 4) A major downside of a fee vs a compensation rank is that it creates a distortion for skilled talent choosing between the US and other countries. If you have the choice between a 300k offer in London vs a 200k offer in New York... 5) The optimal reform effort would be a compensation-ranking that is both age and location-adjusted, with some equity equivalent. Cut off the outsourcers and effectively double the value of the program.(https://t.co/63jIlP7Uit) - **zeeg** (2025-09-21): https://x.com/zeeg/status/1969553801129705536 - All these anti immigration people don’t get it. Trying to block immigrants from “taking jobs” that they are incapable of doing smh Companies will just fully globalize and hire less in the US. We want the best talent, period. Not mediocrity through mandates. - **AILAExecDir** (2025-09-20): https://x.com/AILAExecDir/status/1969548994683547911 - There was no legitimate reason to terrorize H-1B workers and alarm their employers causing 24 hours of panic. The failure to provide a legally sound proclamation and the haphazard messaging caused unnecessary chaos and fear along with human and material cost. https://t.co/NyihbGzA2G - **lennysan** (2025-09-20): https://x.com/lennysan/status/1969491156091486394 - Super underrated book for founders https://t.co/gRlJo1bGE9 - **madiator** (2025-09-20): https://x.com/madiator/status/1969477887943459226 - I love how he has articulated what he wants clearly. But people still don't seem to understand what this is. It's not RAG, it's not notebooklm, it's not lmstudio. It's not a solved problem at all, contrary to what anyone claims. I have also always wanted to something like this, but never got to it due to technical hurdles, but also possibly the business angle. But let's aside the business angle for a minute. Please note that I led a Neurips paper called "Generative Retrieval" (https://t.co/qOT30po0RX) where we transfer the "recommendation knowledge" into a transformer. So that's where my interest in this field comes from. A reasonable start is to do CPT (continual pretraining) of the unstructured text (but needs some data curation) followed by some sort of SFT (and maybe even RL), all of which can be curated. The paper to read here, for the former, is "Synthetic Continual pre-training" from Stanford: https://t.co/D8sPyGNTuy (besides the above one). And it's worth pointing out that ultimately this is first and foremost a data curation problem, and then a modeling problem. The simplest form of what he is saying is to take a pdf (something that's reasonably long) and then train a llm out of it such that it has internalized the knowledge in the pdf and no RAG is needed. Again, this seems easy but it's not. And then of course there is a lot of fascinating technical problems here, besides the challenges around cost (CPT itself is expensive, but the data gen process -- if you read the above paper -- is also expensive). For example, CPT can make the model a bit dumber in other aspects. So how do you methodically add existing pretraining data (which people don't have access to) during the CPT process. - **sulmanrc** (2025-09-20): https://x.com/sulmanrc/status/1969470515388825604 - I went from tn to h1b to citizen. Canada keeps asking why we don’t expand there, meanwhile the US is looking at a 100k fee for h1b. openness built US tech and this is just ridiculous. - **vbingliu** (2025-09-20): https://x.com/vbingliu/status/1969460781495566611 - 🚀 Introducing SWE-Bench Pro — a new benchmark to evaluate LLM coding agents on real, enterprise-grade software engineering tasks. This is the next step beyond SWE-Bench: harder, contamination-resistant, and closer to real-world repos. - **Yuchenj_UW** (2025-09-20): https://x.com/Yuchenj_UW/status/1969446600734068985 - Got flooded with “go back to your country” and “America doesn’t need you” on my H-1B tweet. America is 4.1% of the world. Even if Americans were 10X more likely to be geniuses, 70% would still be born elsewhere. Elon Musk, Jensen Huang, and Sergey Brin weren’t born American. https://t.co/6dzQjXnnPc - **mitchellh** (2025-09-19): https://x.com/mitchellh/status/1969132877104496975 - Based on the past few years of experience, this is the minimum hardware list for desktop app dev: - Multiple monitors of differing DPIs and refresh rates - Multiple keyboards with differing physical layouts - Mouse with both precision/non-precision scrolling - Trackpad that supports gestures - Touchscreen device - Microphone to test dictation inputs - Multiple CPUs, at a minimum x86_64 and aarch64 - Every version of every OS installed and ready for testing (kind of software, but may require extra drives) I wish I could say this was satire, but the matrix of issues to diagnose based on the combination of properties from the above devices is one of the overwhelming challenges of doing desktop application development. - **collision** (2025-09-19): https://x.com/collision/status/1969104042019135950 - Very excited for the next round of Cheeky Pint guests: @destraynor, @pmarca with Charlie Songhurst, @tobi, @RJScaringe, and @CJHandmer. Subscribe to get each episode as it drops. https://t.co/sIgcrGV9Re - **Baconbrix** (2025-09-19): https://x.com/Baconbrix/status/1968903740359925764 - “The more adjectives your product has, the less valuable it is” ~@theo https://t.co/zRuVWthK0j - **embirico** (2025-09-19): https://x.com/embirico/status/1968845953605280043 - Codex changelog: /review We were nervous shipping GPT-5-Codex code review to the @openai team—nobody likes surprise bot comments. Instead it was an instant hit, catching all sorts of crazy bugs. Now you can run it locally in the CLI with /review - **DanielEdrisian** (2025-09-19): https://x.com/DanielEdrisian/status/1968836411773136938 - @nicbstme Correct, this is like a sub-agent. It has a custom prompt + it doesn't "see" the messages of the parent thread, which makes it less biased towards the implementation. It performs better than just asking Codex for a review. - **DanielEdrisian** (2025-09-18): https://x.com/DanielEdrisian/status/1968819243694104899 - Excited to release v1 of /review in the Codex CLI! You can now use gpt-5-codex to investigate and find critical bugs in your code. Check it out and let us know if you have any feedback! https://t.co/pFuI1XJMg6 - **claudeai** (2025-09-17): https://x.com/claudeai/status/1968416788527403371 - In our investigation, we uncovered three separate bugs. They were partly overlapping, making diagnosis even trickier. We've now resolved all three bugs and written a technical report on what happened, which you can find here: https://t.co/taETGOurx0 - **trq212** (2025-09-17): https://x.com/trq212/status/1968405924726587433 - I have a search subagent that can search my inbox, but here’s the twist- Instead of the tool call returning an array of emails, it writes it to a log file and then it uses grep to search across these files. This performed way better and is super easy with the SDK. - **elithrar** (2025-09-17): https://x.com/elithrar/status/1968318992638189912 - amazing to see our very own @ritakozlov on stage with @leinwand at @webflow Conf right now (talking about making things fast, reliable and programmable!) ⚡️ https://t.co/A3qFbMZVVc - **TobinSouth** (2025-09-16): https://x.com/TobinSouth/status/1968101453299613737 - After procrastinating it for half a year, I put my dissertation on arxiv: https://t.co/rLGcnExQ6O - **arampell** (2025-09-16): https://x.com/arampell/status/1968093331067642098 - Cannot emphasize this enough, given how many cool AI things coming out: The battle between every startup and incumbent comes down to whether the startup gets distribution before the incumbent gets innovation. The solution is often to "go boring" - read linked post: - **arampell** (2025-09-16): https://x.com/arampell/status/1968093332363677762 - https://t.co/bYQpagHFaW - **kevinroose** (2025-09-16): https://x.com/kevinroose/status/1968074513553440833 - I don't know how people can look at Waymo's safety data (or just ride in one) and not feel like we live in an age of miracles. The car! Drives! Itself! https://t.co/mzX11DsYDe - **Waymo** (2025-09-16): https://x.com/Waymo/status/1967984942761001026 - All systems go at @flySFO! We’ve been approved by the airport to begin operations, and will start testing soon. More here: https://t.co/0SGUdcl3sj https://t.co/BTfmuhfr8Z - **rohindhar** (2025-09-16): https://x.com/rohindhar/status/1967983307272098140 - This does explain a bit why all the zoning reforms haven’t produced any new housing in California cities like San Francisco One day it’s like “exciting news, we’re now zoned for apartment buildings near mass transit! Build away, yes in my backyard!” Next day is like, “we’ve passed a resolution that landlords are total jerkwads😡” I mean even if it’s true, no one is going to invest in building apartments in that environment 😅 - **arktypeio** (2025-09-16): https://x.com/arktypeio/status/1967981812069675422 - ⚠️trade offer⚠️ i receive: JSON Schema you receive: lightning-fast validation and clear errors https://t.co/aUtqsqxpGs - **KevRGordon** (2025-09-16): https://x.com/KevRGordon/status/1967973073921118262 - Crucial point made by @TheStalwart ... with households now having a record allocation to stocks, it's worth considering how reliant some cohorts of the economy have become on equity markets continuing to do well https://t.co/opGvClGUan https://t.co/nEyFqb8XQu - **EllliotttB** (2025-09-16): https://x.com/EllliotttB/status/1967971596070736009 - Anytime someone tells me they’re investing in apartments in the major blue metros I always think of the days in the life of a thanksgiving turkey chart https://t.co/zdXUP6QhgB - **moseskagan** (2025-09-16): https://x.com/moseskagan/status/1967967894115061850 - @cayimby focused on zoning, but no private investors capitalize building housing under constant threat of not being able to collect rents. - **moseskagan** (2025-09-16): https://x.com/moseskagan/status/1967965133986258976 - LA County once again discussing an eviction moratorium, this time relating to federal deportations of undocumented immigrants. Am begging the Supervisors to internalize the following: 1. Regardless of our personal feelings regarding the fairness or lack thereof of the deportations: Property owners are not social service agencies. Our job is to provide the housing under the terms of the leases we sign with tenants, in exchange for rent. 2. Every time you even discuss eviction moratoriums, it's like hoisting a gigantic red flag telling investors "DO NOT INVEST HERE"... which would be fine, except there is zero chance LA County can build the housing we need with public $$, so we urgently need private investors to provide it. If you would like to help families affected by deportations (a worthy goal!), the appropriate means of doing so is a public program providing targeted rental assistance to those specific families. @kathrynbarger @SupJaniceHahn @LindseyPHorvath @_HollyJMitchell @HildaSolis - **ivanhzhao** (2025-09-16): https://x.com/ivanhzhao/status/1967962858366685320 - Timeless Tools Some tools last a season. Others last a century. From stools to lamps to typewriters, the best designs are humble, adaptable, and made with care. When we named our conference rooms, Notinos voted on their favorite timeless tools. Instead of "Board Room," we now meet in Sony Walkman or Lamy 2000 (one of our fountain pens). We surround ourselves by tools that shape the way we live and work, so we can be inspired and reminded to do the same to software that touches millions people everyday. - **byHeatherLong** (2025-09-16): https://x.com/byHeatherLong/status/1967932733998649658 - This is a key insight into the US economy right now: The bottom 80% of households are just barely keeping their spending up with inflation. Meanwhile, the top 20% are doing just fine and growing their spending. https://t.co/tJpRdTuBEz - **TheStalwart** (2025-09-16): https://x.com/TheStalwart/status/1967901675001352322 - key chart https://t.co/RA3tS4JLzp https://t.co/7EvzhI0UEV - **astuyve** (2025-09-16): https://x.com/astuyve/status/1967806657385451859 - @thdxr Ugh I know, inotify was free during ZIRP - **domenic** (2025-09-16): https://x.com/domenic/status/1967804059676709294 - This seems likely true to me: "the main reason webviews in apps have such a bad reputation is because you don't notice the webviews that are integrated seamlessly." From https://t.co/ZZv2X7ua3h - **cjc** (2025-09-15): https://x.com/cjc/status/1967652743331582380 - You can usually spot this early: the exec who joins and immediately asks for a Chief of Staff, BizOps, or “Strategy” person. It’s often code for I don’t actually want to be hands-on or do the work myself. https://t.co/gNQHOxXFvh - **sam_d_1995** (2025-09-15): https://x.com/sam_d_1995/status/1967648511778017627 - @JMGregorchuk I think the past 50 years of California housing policy does not suggest that prop 13 encourages housing production - **OpenAI** (2025-09-15): https://x.com/OpenAI/status/1967636903165038708 - We’re releasing GPT-5-Codex — a version of GPT-5 further optimized for agentic coding in Codex. Available in the Codex CLI, IDE Extension, web, mobile, and for code reviews in Github. https://t.co/OVGrUovgHN - **clairevo** (2025-09-15): https://x.com/clairevo/status/1967616564397830549 - Let me tell you a dirty secret about a lot of execs: They're extremely smart. And they haven't had to do their own work for years. Look inside any mid->large size company and you'll find VP+ executives that were promoted fast and furious in their early career because they're smart, hard working, make good decisions, have good taste, and can manage up down and sideways well. And as they become more senior, they start to earn the "you're too important to [X]" executive scaffold: - EAs for admin/scheduling/todos - Chief of Staff to keep their directs on track - Sr. leaders working under them eager for opportunity, so take on projects, presentations, meetings, etc. They're still smart, and they're still hard working, and they still make good decisions, so they tend to orchestrate and use these tools at their disposal quite well, choosing what gets done by whom. But - they show up to board meetings with decks made by their team - they show up to sales meetings with prep docs done by the sales person - they share insights generated by some data team - they +@[ea] to schedule every meeting - someone reads & responds to their emails And their job becomes - charm customers - charm candidates - charm the team - charm the board - charm the market - have good ideas (for someone else to do) And before someone shouts "this just optimizes for people who are highly political!" I must emphasize: these people are still usually wicked smart, they're usually extremely charming, and they work really hard (earliest on, latest off.) They hang in the forest, not a tree. Their experience saves your butt once or twice. They just don't have to put their hands on a keyboard and do the things. Sometimes they *can't* put their hands on a keyboard and do the things, because they're in endless meetings and on endless trips to do the charming/idea things. But after awhile of this, their "doing the things" muscle atrophies. Your CMO can't write compelling copy. Your CPO doesn't look at designs anymore. Your CRO can't login to demo. Your VPE doesn't have the latest local env setup. And over time, as an exec, your ability to be wicked smart degrades with your distance from the work, especially when things like AI come and smack your team in the face. This is all to say, I have two warnings for you if you've made it this far: EXECS: Do not stop doing the things. Take on projects, write your docs, do your own damn analysis, and don't stop touching the work. Fix your calendar so you're not just bopping meeting to meeting, and use that big brain of yours do actually build something. TEAMS: Consider yourself lucky if you have a leadership team that can still/wants to do the things. Trust: you'd rather have a micromanagey CEO who drops suggested edits in your doc than a manager-class exec team that doesn't even know your doc exists. Your company will be better for the work, the specificity, the care that comes with a doing-the-things leadership team, than a organizes-the-work leadership team. And for all of you: AI will generate a little microcosm of the dynamic above, but for IC work. I love AI, but still think it's important to exercise the muscle daily of writing, coding, reading, speaking, thinking. All unused skills will atrophy. Make sure you get stronger, not weaker, with these new tools. Ok back to doing things 😎 - **Scott_Wiener** (2025-09-15): https://x.com/Scott_Wiener/status/1967591881447252023 - That “legacy” is heavy-handed political control over everything that happens in the district: micro-managing small businesses, making them beg for permits & crushing new housing production. I’m proud of Supervisor @DannySauter for challenging this broken & corrupt status quo. https://t.co/q1xA4EXsZx - **TheStalwart** (2025-09-15): https://x.com/TheStalwart/status/1967522089742811161 - NEW ODD LOTS: The Cardboardboxpocalypse @tracyalloway and I talked to Ryan Fox, the containers and packaging analyst for Bloomberg Intellligece about the decline in US cardboard box volumes, and what that indicator says about the economy. https://t.co/wGWuLYRj5R - **cjc** (2025-09-14): https://x.com/cjc/status/1967333674128691630 - Working Hours, Then and Now The debate around the 9-9-6 work schedule strikes me as oddly shallow. Startups in China have been doing this for years, and I remember back in the mid-2010s it was practically an article of faith among people watching the industry that this was why Chinese startups would beat their American counterparts. But here’s the truth: no one has to force you to work long hours if you really want to. The founders and early employees who push hardest aren’t usually doing it because someone set the office lights to stay on until midnight. They’re doing it because they care. It’s not obedience—it’s compulsion. The Role of Urgency In industries where being first really matters—ridesharing a decade ago, or AI today—the only advantage you can really compound is speed. If you and your competitors are equally talented, and you both see the same opportunity, the one who simply spends more hours in the problem space has the edge. Time becomes the rawest form of capital. This doesn’t mean it’s sustainable forever. It rarely is. But there’s a reason why some companies choose it anyway: because the window for success may only open once. The Shift That Happens With Age When I was younger, I worked longer hours, and not because someone told me to. At 22, few other time-intensive priorities, my work was my best use of time. But as I’ve gotten older, I’ve realized that companies aren’t just paying for hours anymore. They’re paying for judgment. For the ability to make good calls quickly. For the ability to know which doors are worth opening. They’re paying for experience, for the network that makes introductions possible, and for the intuition that keeps you from wasting months chasing the wrong idea. So yes, I work fewer hours now. But I don’t feel guilty about it. If anything, the work I do now counts for more than the hours I used to put in. Time as Leverage There’s a progression here that mirrors how startups themselves evolve. In the earliest stage, what you need is sheer effort. Later on, effort compounds into something more valuable: leverage. A founder who is good at hiring can accomplish more in a 50-hour week than they used to in a 90-hour one, because they’ve figured out how to multiply their time. Individuals work the same way. What you do in your early 20s is different from what you do in your 30s or 40s, because what you’re bringing to the table is different. If you still think output is proportional to hours worked, you’re probably underrating the compound returns of experience. And that’s why I don’t worry about not grinding quite as hard as I used to. The form of work has changed, but not the commitment to it. - **southpolesteve** (2025-09-14): https://x.com/southpolesteve/status/1967217802672328788 - The breakfast spot near me serves two small burritos when you order a “burrito.” I like to think this has massively impacted the net happiness of the neighborhood. Everyone is walking around looking to share their second burrito with someone. - **swyx** (2025-09-14): https://x.com/swyx/status/1967112272217862461 - periodic reminder to just invert everything u think u know when doing developer marketing https://t.co/CqkPI1UBQL https://t.co/bRZj2oJX0E - **zephyr_z9** (2025-09-13): https://x.com/zephyr_z9/status/1967014180533276829 - > be Google in 2017 > small team drops “Attention Is All You Need” on arXiv > execs nod politely, go back to selling ads for socks > let Transformer gather dust for 5 yrs like a vintage Beanie Baby > be Noam Shazeer, OG wizard > quits, builds AI-boyfriend app “https://t.co/KRAcLptABl” > millions of lonely hearts pay $9.99 to e-date 2B from Nier > Google HR: “any chance you wanna come back?” > Shazeer: “busy, my e-gf is calling” > 2022, OpenAI drops ChatGPT > world loses its mind, Google stock does bungee jump > Sundar holds all-hands: “guys, we need a response by Q2” > team ships Bard in 3 weeks using 12 interns, 2000 TPUv4s, and pure panic > Bard hallucinates that the earth is flat > Wall Street: “sell sell sell” > Sergey hears the screams from his yacht > flies back on a carbon-negative jet, reclaims unofficial crown > merges GoogleBrain + DeepMind like Dragon Ball fusion > axes Bard, slaps Gemini sticker on everything > Gemini launches, immediately generate a black Abraham Lincoln > Twitter dubs them “Bing’s woke cousin” > Shazeer finally answers HR’s 47th email >“fine, but I want my own parking spot and unlimited LaCroix” > returns, waves wand, Gemini 2 & 2.5 drop > context window so big it can summarise the Bible + your group chat > Google processing 980T tokens/month, equal to 40 % of human SMS traffic > someone sneezes “Nano Banana” into the Play Store > overnight it’s #1 app, turns ur selfie into big-tiddy anime waifu - **friscolive415** (2025-09-13): https://x.com/friscolive415/status/1966923935754449050 - ⚾️🦭No one in the entire stadium felt that walk-off grand slam coming like Lou Seal did. He even took his sunglasses off so he could experience it in all its glory. https://t.co/L40KMvRlaJ - **ysl_serena** (2025-09-13): https://x.com/ysl_serena/status/1966923646209061291 - The amount of alpha you get by just being in SF 🥹 - **thdxr** (2025-09-13): https://x.com/thdxr/status/1966915837308399751 - this is the right way to do it https://t.co/ONVSPINe3G - **Padday** (2025-09-13): https://x.com/Padday/status/1966914649011728583 - Honestly another big problem is that many marketers don't really understand the technology they are marketing. In Saas you could get away with it. In AI you absolutely cannot. Marketers are going to need to get way more technical to succeed. But many don't want to because they just aren't into it enough. - **Padday** (2025-09-13): https://x.com/Padday/status/1966910076662259747 - Couldn't agree more. I've been @intercom CPO for over a decade but am now running marketing because we need a different set of skills to win. I think a major part of the problem is that a whole generation of marketers have learned and worked with a specific playbook for a decade and they can't seem to unlearn it. That playbook included abstracting away the technology and obsessing over 'leading with the customer benefit', which then leads to boring, bland, marketing. The new way needs to blend: - customer problem - exciting new technology capability - education via laddering messaging from simple to complex - **aidenybai** (2025-09-13): https://x.com/aidenybai/status/1966905966861164996 - iframes are an incredibly web feature - **sama** (2025-09-13): https://x.com/sama/status/1966899254804574266 - i suspect society was better off with phone call culture than meeting culture - **zephyr_z9** (2025-09-13): https://x.com/zephyr_z9/status/1966800432380141647 - > be American soyboi > plant 80 million acres of beans > check futures: "literally gold with pods" > ping China: “usual 40 million tons?” > China: “lol tariffs go brrrr, we buying from Brazil now” > try selling to vegans: they only want organic non-GMO sadness > offer free soy latte with every 10 ton purchase, still no takers > Congress: "just eat more tofu bro" > farmers: "sir this is Iowa" > mfw Trump tweets "trade wars are good & easy" while my field turns into Minecraft jungle - **nizzyabi** (2025-09-13): https://x.com/nizzyabi/status/1966779773809463389 - never understood the “this app is just a ___ wrapper” argument everything is a wrapper if you think about it the whole point of a company is to make something more convenient - **hanlonbt** (2025-09-12): https://x.com/hanlonbt/status/1966629766879998017 - After 8 years, we did it! SB 79, the biggest upzoning bill in state history, just passed the Legislature. Brief 🧵, with some important thank yous. 1/x https://t.co/kjSpT7cq5t - **mnolangray** (2025-09-12): https://x.com/mnolangray/status/1966629139873415611 - SB 79 HAS PASSED! https://t.co/9p4eSAWCFj - **akothari** (2025-09-12): https://x.com/akothari/status/1966624391770316844 - No opinion on 9-9-6. Strong opinion on 7–8 hours of sleep. That's the ultimate productivity hack! - **thdxr** (2025-09-12): https://x.com/thdxr/status/1966623643989143575 - readable code is simple - you have primary functions that compose together a bunch of smaller functions when i wrote elixir this was natural to do i've realized after years in ts my habits eroded for the following reasons 1. extracting out logic into a function is more overhead because you have to write out argument types. sometimes these are complex inferred types you wouldn't have to articulate when logic is inline 2. lack of pipe operator - having a lot of small functions is less fun when you need to name intermediary variables to name every step 3. camel case - i hate camel case and small functions usually can't be one word. `checkUsage` vs `check_usage` trying to be more aware of this because it's biasing me towards fat unreadable functions - **saltyAom** (2025-09-12): https://x.com/saltyAom/status/1966613110086922461 - @irvinebroque Libraries that use this feature: at least one, Elysia If you put a value instead of a function, it gets converted to a Response in memory before the server starts When needed, it clone the Response instead of creating a new one It's beneficial for static files like a large video https://t.co/nbV8xPtLgT - **saltyAom** (2025-09-12): https://x.com/saltyAom/status/1966611498207506615 - @irvinebroque Not Request, it’s a Response The main reason is performance Creating a new Response is more expensive than reuse the existing one by cloning Use case: caching, static content like image or static HTML, error handling that always return the same result like 404, 502 and so on - **thdxr** (2025-09-12): https://x.com/thdxr/status/1966610577482031496 - laravels design has become so good i always see david's posts and get really jealous https://t.co/YyAwPWDH6r - **StainlessAPI** (2025-09-12): https://x.com/StainlessAPI/status/1966600524238594083 - What do robust and polished SDKs look like? They handle scale, network errors, and telemetry without breaking, and surface the right details in your IDE: inline docs, type hints, contextual errors. @RattrayAlex dives into what it means to obsess over quality at Stainless. https://t.co/cWW8XmTlz4 - **jmwind** (2025-09-12): https://x.com/jmwind/status/1966595322224013562 - The biggest drain on your team isn't the impossible challenges. It's the people who turn simple tasks into complex ones. Keep the simplifiers. Avoid the complicators. It's not easy to spot, often the complicators are seen as having a great attention to detail & high care. But it's not that. It's inability to self-regulate on the right detail, the right problems, and the right effort. And their blast radius is hard to contain. - **kendra__andrews** (2025-09-12): https://x.com/kendra__andrews/status/1966592250600714514 - Welcome to ✨Ballhalla✨, the home of Golden State Valkyries that has quickly become one of the best home courts the league. My ESPN video essay on the frenzied fandom in the Bay Area that helped propel the Valkyries into their first postseason: https://t.co/DwGc83YqxL - **Baconbrix** (2025-09-12): https://x.com/Baconbrix/status/1966579358895861982 - Every Expo app generator just gets these APIs for free today 🚀 Huge native quality increase for users of Rork, Bolt, Bloom, Replit, vibecode, and many more… https://t.co/bulioHIfot - **whoiskatrin** (2025-09-12): https://x.com/whoiskatrin/status/1966546273814602061 - shipping never stops at Cloudflare we've just added support for icons, links etc in Agents SDK, so you can make your MCP servers really pretty 🧡 https://t.co/4TzZv1IaEG https://t.co/vrnjUQn48O - **LakeAustinBlvd** (2025-09-12): https://x.com/LakeAustinBlvd/status/1966536611258130632 - We're hiring a design engineer at Cloudflare to work on some of the most high-visibility, high-impact resources we have (our Developer Docs, our Radar tool, our Community forum and more). DMs open. Come help build the future of how people learn about and use the Internet. - **ritakozlov** (2025-09-12): https://x.com/ritakozlov/status/1966533562490126626 - it's friday! continuing this thread with more @cloudflaredev people you should follow: @darkmembo — knows basically everything about how devplat works @burcs @nandafyi — follow for cloudflaredev + vibes @thomasgauvin — is putting himself out there (!). great demos, tutorials, and devplat content - **ritakozlov** (2025-09-12): https://x.com/ritakozlov/status/1966528047063978030 - my favorite part of the @cloudflaredev weekly demos is seeing all the incredible features / products / improvements we're about to launch my second favorite part is seeing how many and which tabs different people have open. tells you a lot about a person - **tommoor** (2025-09-12): https://x.com/tommoor/status/1966515265362862163 - Just had a 1-1 meeting where we identified a bug while using @linear and filed an issue – it was triaged, fixed, and deployed to production before the call ended. NICE - **arjunattam** (2025-09-12): https://x.com/arjunattam/status/1966383790617763986 - i just <3 cloudflare was struggling to set up dashboards on AWS to track some metrics yesterday took an hour to send everything to cloudflare workers analytics engine + built a custom ui with an llm custom dashboard purpose-built for my exact problem -- done ✅ - **kendra__andrews** (2025-09-12): https://x.com/kendra__andrews/status/1966347329688248725 - “Napheesa Collier is the most unguardable player in the league.” ESPN story on Collier clinching the elusive 50/40/90 shooting split in her regular-season finale and final move of her MVP campaign: https://t.co/ELcug3Bt70 - **destraynor** (2025-09-12): https://x.com/destraynor/status/1966326997560205812 - Tonight is our first deeply technical AI event in quite some time, to share some of our deep and difficult learning building @Fin_ai. We also have published a new site and a lot of new articles, links in thread https://t.co/iPeesre9Sw - **kendra__andrews** (2025-09-12): https://x.com/kendra__andrews/status/1966317596107747362 - Napheesa Collier finishes the regular season shooting 53% from the floor, 40.3% from three and 91% from the free thrown line. She joins Elena Delle Donne (in 2019) as the only qualified players in WNBA history to have a 50/40/90 season. - **zebassembly** (2025-09-12): https://x.com/zebassembly/status/1966295870309347558 - @JFSIII I wonder what that zebp guy is cooking up 🤔🤔🤔 - **Baconbrix** (2025-09-11): https://x.com/Baconbrix/status/1966241822591930461 - You can now launch an Expo website in less than a minute. We are simultaneously speeding up EAS and the Expo to accomplish this same feat on native. https://t.co/AiizDBf0ba - **CloudflareDev** (2025-09-11): https://x.com/CloudflareDev/status/1966220511224250856 - Our SF office was filled with developers from all over the world, good vibes, and good conversations last night for MCP night w/ @jamdotdev! https://t.co/cQyX3IEPSV - **zatlyn** (2025-09-11): https://x.com/zatlyn/status/1966199995817926915 - Something fun is brewing at @Cloudflare SF today…excited to launch it soon cc @craigsdennis https://t.co/ZAFfAyp5T2 - **JFSIII** (2025-09-11): https://x.com/JFSIII/status/1966167836059107526 - White Whale captured?! https://t.co/lzLNNNxL5q https://t.co/aGlZf5ucnJ https://t.co/SCvclVugY2 - **kentcdodds** (2025-09-11): https://x.com/kentcdodds/status/1966140010434015456 - This is what I'm doing and what I'll teach you how to do as well. Cloudflare has the technology to get your services in front of your users in the AI agent they prefer and the time to learn how is now. https://t.co/lnZL5Nxt9s - **zephyr_z9** (2025-09-11): https://x.com/zephyr_z9/status/1966138284100432131 - > be me, NVIDIA in 2006 > making GPUs so gamers can play World of Warcraft in HD > Jensen Huang has a vision > "what if we made them do... math?" > entire boardroom thinks he's finally lost it > create CUDA > a software prison so elegant, academics will volunteer to be inmates > gamers are confused. "Will this make Crysis run faster?" > Jensen: "...no :)" > fast forward to 2012 > AI researchers, hopped up on adderall and free pizza, discover they need a shitload of matrix multiplication to make cat pictures > their Intel Xeons are crying, smoking, and filing for divorce > they discover CUDA > "holy shit this math thing you made is perfect" > the entire field of AI becomes a subsidiary of a company known for making Fortnite run better > we see this and our eyes turn into dollar signs > decide the prison needs better amenities > invent Tensor Cores > "what if part of the chip did ONLY the AI math, but did it at ludicrous speed?" > it's like adding a nitro booster to a Honda Civic specifically for going to the grocery store > AMD's competing card is over there trying to do the same math with a abacus and hope > because researchers are so lazy they can't be trusted to write a for-loop > just give them a big red "MAKE AI" button that runs our code > create cuDNN > bake it directly into TensorFlow and PyTorch > ecosystem lock-in is complete. We own the land, the factory, and the souls of the workers. > a bunch of hippies from a thing called "OpenAI" call > they want to build AGI or something > they need compute > roll up to their office, which is probably a converted garage, with the world's first DGX-1 > it's like the monolith from Space Odyssey > I tell them it has 8 GPUs connected by NVLink, so they can gossip like schoolgirls at light speed > they ask what it does > "It does AI, idiots. Just plug it in." > they sign the papers. > But we're not done. The prison needs a better yard. > See them trying to connect our beautiful DGX boxes with CAT-5 cables > embarrassing. > Acquire Mellanox for their InfiniBand tech > Now the nervous system connecting the GPUs is also ours > NVSwitch makes an entire rack of GPUs hold hands so tight they become one colossal mega-brain > Meanwhile, in a dimly lit basement in Santa Clara... AMD > They drop a new GPU. "The AMD Instinct MI300X! We have more theoretical FLOPs!" > "Better Benchmark performance." > One brave researcher tries to use it. > spends 6 weeks trying to install ROCm, their knock-off https://t.co/c336ngfjCA version of CUDA > gets 400 errors, a kernel panic, and divorce papers from his wife > he finally gets a model to run at half the promised speed > The benchmark was for a matrix size of 1x1 > He throws the card in the trash and buys an NVIDIA H100 on the company card. The cost is irrelevant. > Sanity is priceless. > Jensen takes the stage. The leather jacket has its own gravitational pull. Unleashes the GB200 NVL72 > 72 GPUs and 36 Grace CPUs in a single NVLink domain with 130 TB/s of bandwidth > the GPUs share thoughts before they even have them. > 2025 > AMD comes out with their "rack scale" solution, the MI355X 128-GPU rack > It's just 16 groups of 8 GPUs duct-taped together with Ethernet > A scale out clown car > We announce Rubin CPX GPU enabling cheaper prefill during inference > saving $$$$ > it's the ultimate vendor lock-in > every other chip designer sees this and has to throw their architecture in the trash and go back to the drawing board > All hail the leather-clad prophet. - **bhalligan** (2025-09-11): https://x.com/bhalligan/status/1966135730268496053 - There are archetypes that I find hard to work with: The clarifier: you can never give them something relatively simple to do. Always a barrage of clarifying questions. The finger pointer: it’s always someone else’s “fault.” - **darioperkins** (2025-09-11): https://x.com/darioperkins/status/1966125897674592502 - cpi is hot the jobs market is about to stall whoever invented these tariffs is a bit of a fool - **thdxr** (2025-09-11): https://x.com/thdxr/status/1965958159207641178 - build primitives then use them to create a good out of the box experience advanced people can go deeper there's no trade off between simplicity and customization - it's just harder to do - **yagiznizipli** (2025-09-10): https://x.com/yagiznizipli/status/1965927084687286609 - It’s been a productive couple of days with James https://t.co/cQeUDtsVhT - **dsp_** (2025-09-10): https://x.com/dsp_/status/1965914625578520798 - Great to see more MCP adoption in the world. Our little project at Anthropic has come far. https://t.co/NLEJijvS3p - **CAgovernor** (2025-09-10): https://x.com/CAgovernor/status/1965899170579202144 - We should all feel a deep sense of grief and outrage at the terrible violence that took place in Utah today. Charlie Kirk's murder is sick and reprehensible, and our thoughts are with his family, children, and loved ones. I knew Charlie, and I admired his passion and commitment to debate. His senseless murder is a reminder of how important it is for all of us, across the political spectrum, to foster genuine discourse on issues that deeply affect us all without resorting to political violence. The best way to honor Charlie's memory is to continue his work: engage with each other, across ideology, through spirited discourse. In a democracy, ideas are tested through words and good-faith debate — never through violence. Honest disagreement makes us stronger; violence only drives us further apart and corrodes the values at the heart of this nation. - **ritakozlov** (2025-09-10): https://x.com/ritakozlov/status/1965875081206505881 - TIL "write once, run anywhere" also came from sun microsystems (in addition to "the network is the computer") wow that company was full of banger slogans that are now perfect for cloudflare. truly ahead of their time - **elithrar** (2025-09-10): https://x.com/elithrar/status/1965784586966933683 - If you're a seasoned, staff or principal level engineer with a background in distributed systems, C++ and/or databases and you want to come and work on the next version of the routing & metadata layer that powers Durable Objects — DM me. Big role for the right person. - **whoiskatrin** (2025-09-10): https://x.com/whoiskatrin/status/1965776940129829169 - 🎈 we just shipped agents SDK v0.1.0 and workers-ai-provider v2.0.0 with AI SDK v5 support - comes with out of the box automated migration - useAgentChat hook can now automatically detect which tools require confirmation - workers AI models now support v5 file handling with automatic conversion: we've added some docs and guides (more coming soon!). go build cool stuff! - **ritakozlov** (2025-09-10): https://x.com/ritakozlov/status/1965769174384619994 - working on something i’m nervous about it yall be nice https://t.co/RYHQVdg0eT - **samgoodwin89** (2025-09-10): https://x.com/samgoodwin89/status/1965611269635146203 - Finally got Monorepos working in @alchemy_run Here I `dev`, `deploy` and `destroy` a Backend Worker that is imported and bound to the Frontend. https://t.co/9EDdi6TIrA https://t.co/C5pzYWH9Kr - **thdxr** (2025-09-09): https://x.com/thdxr/status/1965552383184044140 - opened up bluesky for the first time in a while was reminded of a lot of people who's posts i used to enjoy are now entirely posting there kinda sad that that this whole fracture happened - **thdxr** (2025-09-09): https://x.com/thdxr/status/1965533159254262209 - being performative about work will forever be a thing it always manifests in some new way no one getting things done is counting their hours, whether it's high or low https://t.co/rCzNWPTQFM - **mattzcarey** (2025-09-09): https://x.com/mattzcarey/status/1965502051364077853 - Introducing the first open source MCP Tool Testing Agent powered by Claude Code >be me >build the most useful MCP server known to humankind >description pristine, all the arguments you could want, >ship it, go outside, stare at cloud >“just ask your model to use the tools,” >claude: “what if… I don’t” >gpt: “I have called mcp__my_server_easy_tool with parameter blorpo as a stringified array” >blorpo not in schema >users sad >me sadder plot twist npx @ stackone/mcp-test --transport http --url https://amazing-server-url .com/mcp point it at your server, it handshakes, enumerates tools, pokes endpoints with fun data. So.... Can Claude really use your MCP Server? PRs welcome, stars cure my impostor syndrome :) - **jarredsumner** (2025-09-09): https://x.com/jarredsumner/status/1965489392795172878 - When a < 100 LOC npm package gets tens of millions of monthly downloads, that’s a hint that it should perhaps be part of the runtime. - **thdxr** (2025-09-09): https://x.com/thdxr/status/1965467435274018950 - LLM coding magnifies your judgement it's weird to assume they only magnify your good tendencies it's weird to assume they only magnify your bad tendencies - **yomnashousha** (2025-09-09): https://x.com/yomnashousha/status/1965426715255341210 - Wrangler now has zero-config static deployments 🎉 Before: Long command with multiple flags After: `wrangler deploy [dir]` That's it. Interactive prompts handle everything else. https://t.co/oPexG8HWK7 - **ValDotTown** (2025-09-09): https://x.com/ValDotTown/status/1965417600982405317 - this summer @_404Wolf rebuilt our TypeScript editor tooling from scratch, building on @deno_land's LSP server, @Cloudflare Containers. it's way better now. it was a huge project. read how he did it: https://t.co/w3B4VhrCDp and all the code is open source! - **jamonholmgren** (2025-09-09): https://x.com/jamonholmgren/status/1965273941456683456 - I’m warming up to this wild idea of checking node_modules into git. Think about it … a version bump doesn’t just update one line. You see a diff of all the code that changed with that version bump, and can review it too. After all, it’s your code now. You’re running it. https://t.co/6CZA57kBE8 - **htmx_org** (2025-09-09): https://x.com/htmx_org/status/1965247887773954227 - resolving dependencies at deployment time is literally the most insane thing this industry has ever normalized and that is saying something given how psychotic this industry is let me tell you bub - **mjackson** (2025-09-09): https://x.com/mjackson/status/1965243804811919864 - Remix 3 https://t.co/wLiaVvCIYc - **thdxr** (2025-09-08): https://x.com/thdxr/status/1965143038688473112 - when we needed claude inference capacity we went straight to google didn't even bother with anthropic makes you think about how this ecosystem will shake out - **mattzcarey** (2025-09-08): https://x.com/mattzcarey/status/1965064402388005354 - @zeror404 @Cloudflare @ghostwriternr does. - **StainlessAPI** (2025-09-08): https://x.com/StainlessAPI/status/1965055827175469087 - API pain isn’t in the code, it’s in the design. The right split: humans set intent, AI writes the middle layer, the platform handles infra. Stainless is that platform. From @RattrayAlex's conversation with @JenniferHli on the @a16z podcast. https://t.co/IACJPpuudk - **KentonVarda** (2025-09-08): https://x.com/KentonVarda/status/1965048470114640168 - Every time I open devtools it's trying to tell me what's new and every time I'm just there to do some basic arithmetic in the console or to edit a page element that's annoying me, just like I've been doing for decades. - **ritakozlov** (2025-09-08): https://x.com/ritakozlov/status/1965041869513269279 - run express.js on cloudflare workers 🚀 https://t.co/RYiD8N0svZ https://t.co/6yMjggrtK6 - **Cloudflare** (2025-09-08): https://x.com/Cloudflare/status/1965038799102116006 - We've implemented the node:http client and server APIs in Cloudflare Workers, allowing developers to migrate existing Node.js applications with minimal code changes. https://t.co/RHP57jUlqe - **cramforce** (2025-09-07): https://x.com/cramforce/status/1964776855174889863 - Just imagine the damage that is being done to our country for literally no gain whatsoever https://t.co/RNruniQ1hz - **BraydenWilmoth** (2025-09-07): https://x.com/BraydenWilmoth/status/1964763906012049628 - @burcs give me html, javascript and css. render csr for dynamic experiences (dashboards) render ssr for static sites (marketing) add aria-label's to make elements accessible. use good color contrast. web dev is easy. - **burcs** (2025-09-07): https://x.com/burcs/status/1964762907297886451 - sse vs csr micro vs monolith spa vs mpa only two things matter to your users: does it look good? is it fast? - **burcs** (2025-09-07): https://x.com/burcs/status/1964761692656771222 - frontend should be easy, for whatever reason people dedicate their entire careers complicating it... - **ritakozlov** (2025-09-07): https://x.com/ritakozlov/status/1964727674183844094 - when you start writing, there’s nothing more painful than scrapping everything and starting over — it feels like such a waste as you get better, you realize it’s part of the process and not a waste because it takes each iteration for your thinking getting sharper - **waterdoggie** (2025-09-07): https://x.com/waterdoggie/status/1964672020177986028 - @__morse @framer Use the hibernating web sockets api - **CloudflareDev** (2025-09-06): https://x.com/CloudflareDev/status/1964454610544009505 - Want to run an Iceberg data catalog on top of R2 and connect your favorite query engine (like DuckDB or Trino) - ? 👇 https://t.co/ZJoNyFNYD2 - **mwylde** (2025-09-06): https://x.com/mwylde/status/1964427450500436370 - {r2_warehouse} 👀 https://t.co/vqPa9FLYzU - **thdxr** (2025-09-06): https://x.com/thdxr/status/1964361553429319748 - it's funny to be protesting a company because you fell for their marketing and believe in them harder than anyone https://t.co/UugfT7Nqwp - **ankrgyl** (2025-09-06): https://x.com/ankrgyl/status/1964356420205269104 - for me at least, the one thing python really has over typescript is the REPL. I dream and hope for something like the browser console in my terminal. Someday... - **just_some_dev** (2025-09-06): https://x.com/just_some_dev/status/1964336388007993831 - @irvinebroque @dok2001 @Cloudflare JSON lol, our build creates some massive JSON blobs with data we scrape for https://t.co/Uxc7GdHI48. We don’t use it for the site itself, but people rely on it as a sort of API. - **just_some_dev** (2025-09-06): https://x.com/just_some_dev/status/1964336016212312341 - @irvinebroque @dok2001 @Cloudflare Well the problem is the file size limit more than the item count. A way for wrangler to check if it uploaded things before and tier uploads to r2 or something would solve this. I wrote ts to do this with kv and r2 and it works, I just wish wrangler did it. - **MattieTK** (2025-09-06): https://x.com/MattieTK/status/1964255627103895593 - @heidar_an @dok2001 @Cloudflare You absolutely can PR docs, which are the ones you would think need the most work? - **aninibread** (2025-09-06): https://x.com/aninibread/status/1964172489249976455 - What should we name her? 💾 https://t.co/oiioG5DIfc - **brandinpodziem2** (2025-09-06): https://x.com/brandinpodziem2/status/1964155023849340940 - Natalie Nakase…. Not even close 🤝🏼 https://t.co/YU9dO0gLQI - **joannejang** (2025-09-05): https://x.com/joannejang/status/1964107648296767820 - 🧪 i’m starting oai labs: a research-driven group focused on inventing and prototyping new interfaces for how people collaborate with ai. i’m excited to explore patterns that move us beyond chat or even agents -- toward new paradigms and instruments for thinking, making, playing, learning, connecting, doing. this builds on the work i’ve had the most fun doing over the past 4 years at openai: taking frontier capabilities and crafting them for the world with brilliant coworkers. from dall·e 2 and standard voice mode to gpt-4 and the model behavior discipline -- spanning personalities and personalization -- i've learned how much the shape of an interface can invite people to push the edges of what feels possible. for this effort we'll optimize for exploration over polish, so i'm excited to bring some very bad ideas & prototypes to life!! - **skeptic_fx** (2025-09-05): https://x.com/skeptic_fx/status/1964036205437473257 - IKEA website felt faster than it used to be. Then I checked cdn-cgi/trace. ofc. - **dok2001** (2025-09-05): https://x.com/dok2001/status/1964031296604147955 - If one of your angel investors berates you into using their product, they aren't going to be there for you when times are hard. The first thing I tell founders I invest in is: always do what's best for your business, get advice, but go with your own gut. - **michellechen** (2025-09-05): https://x.com/michellechen/status/1964016960569557309 - the dolomites https://t.co/pFjbO6qUlc - **samlambert** (2025-09-05): https://x.com/samlambert/status/1963995948755157089 - why would you knowingly choose the slow database? https://t.co/dYYwBrBmW9 - **KentonVarda** (2025-09-05): https://x.com/KentonVarda/status/1963966469148180839 - After struggling for years trying to figure out why people think Durable Objects are complicated, I'm increasingly convinced that it's just that they *sound* complicated. Feels like we can solve 90% of it by renaming `DurableObject` to `StatefulWorker`? - **michellechen** (2025-09-05): https://x.com/michellechen/status/1963957544554041620 - @ritakozlov @elithrar @irvinebroque @korinne_dev GREMLIN ??!! - **ritakozlov** (2025-09-05): https://x.com/ritakozlov/status/1963955054911656311 - got lots of love yesterday, so naturally need to pay it forward. which is not hard because i work with truly the best actually going to make this a friday tradition bc i work with many excellent people anyway, here are some incredible people you should follow 👇 https://t.co/e1ldI01XgK - **ritakozlov** (2025-09-05): https://x.com/ritakozlov/status/1963955057109381420 - @elithrar — can't build a compute platform without its best friend storage (plus, bonus car content) @irvinebroque — literally the person that makes things on workers happen @michellechen — ai gremlin @korinne_dev — making workers QoL go 📈 by the day - **IamJerm24** (2025-09-05): https://x.com/IamJerm24/status/1963836385435226447 - @kendra__andrews Why not just play in Oakland where their headquarters is? Oracle is right there. - **Nichols_NTA** (2025-09-05): https://x.com/Nichols_NTA/status/1963825949616885963 - @kendra__andrews Not using Oracle is CRAZY - **thougy** (2025-09-05): https://x.com/thougy/status/1963825136810999844 - @kendra__andrews This is a huge missed opportunity. Had they decided to go instead to Oracle in Oakland, they could have called it Thor-acle. Missed out on a great tie-in to the Valkyries-Ballhalla theme. - **JepsenZander** (2025-09-05): https://x.com/JepsenZander/status/1963823043169439817 - @kendra__andrews Someone needs to explain why it’s at SAP center instead of Oracle - **ezlolo** (2025-09-05): https://x.com/ezlolo/status/1963822119437144531 - @kendra__andrews Not even Oracle - **jainavgohel** (2025-09-05): https://x.com/jainavgohel/status/1963821141438644503 - @kendra__andrews Dang, I wish it was at Oracle Arena - **zeeg** (2025-09-05): https://x.com/zeeg/status/1963814682097274931 - I think the fundamental issue is the evals are simply too unreliable and not scientific enough. Yes you could build crazy statistical test suite but iteration is a nightmare because don’t have a clue why shit fails. This is a real issue with LLM-based dev imo. https://t.co/aTezXJzY6S - **kevinweil** (2025-09-05): https://x.com/kevinweil/status/1963812884560920788 - 🎉 https://t.co/EBmmu2YEuA - **swyx** (2025-09-05): https://x.com/swyx/status/1963795147852005516 - just remembered this banger https://t.co/bYHIGHEPTE - **kentcdodds** (2025-09-05): https://x.com/kentcdodds/status/1963772523067093216 - https://t.co/5i1GVbKTPf https://t.co/9aFXVOAi5k - **ankrgyl** (2025-09-04): https://x.com/ankrgyl/status/1963730689876185297 - There's actually a surprising and interesting nuance in your point -- We too have noticed that coding companies do far less evaluation than products in other areas. There are many possible explanations for this, which may be fun for us to talk about live. - **clattner_llvm** (2025-09-04): https://x.com/clattner_llvm/status/1963715762876723614 - It was wonderful to get to discuss GPU programming and architecture with Ron. We explored why new technologies are useful: CUDA is 20 years old, and Python and C++ are from the 80’s. It is time for new ideas and a new approach! 🔥 https://t.co/ne9gbQecou - **ritakozlov** (2025-09-04): https://x.com/ritakozlov/status/1963705919922987268 - OH in a meeting: “historically when you deployed an mcp server” - **ImSh4yy** (2025-09-04): https://x.com/ImSh4yy/status/1963690160387723472 - Yeah, Cloudflare is good. - **ritakozlov** (2025-09-04): https://x.com/ritakozlov/status/1963676847167983879 - another change that's been a long time coming! we've increased the limit for static assets in @cloudflare workers from 20K to 100K files 🔥 https://t.co/0nE7loa8bD - **ritakozlov** (2025-09-04): https://x.com/ritakozlov/status/1963675095291056228 - really proud of the team's work on this we shipped many new features over the years which has caused the workers api to bloat, so it was time to simplify (and you can literally see the difference) these changes enable much better SDKs and terraform support too https://t.co/ImBTG7IuDQ - **blankparticle** (2025-09-04): https://x.com/blankparticle/status/1963674401276354703 - @korinne_dev @CloudflareDev can we have some better errors 🙏 there is no way to know what error does this API throw if an worker with same name already exists https://t.co/z8BgPD0COg - **korinne_dev** (2025-09-04): https://x.com/korinne_dev/status/1963672649160433817 - New REST API for deploying @CloudflareDev Workers just dropped before vs after https://t.co/NXi4PPDNoA - **dok2001** (2025-09-04): https://x.com/dok2001/status/1963622798158430321 - I have spent 14 years helping to build a better Internet @Cloudflare. I say helping because no company can do it alone. Real progress comes from working with others who share this spirit. - **KentonVarda** (2025-09-04): https://x.com/KentonVarda/status/1963622596722704636 - Me: Sunil, I think everyone using MCP wrong. What if they did [new approach], enabled by [unannounced new Workers feature]? Sunil: Makes sense, I'll add support for that to agents-sdk tomorrow and we can try it out. https://t.co/L1H5oR5Wyr - **dok2001** (2025-09-04): https://x.com/dok2001/status/1963611108998095246 - Triangle Man is a decent front end designer, but he knows nothing about networks. He keeps having to make this point because he has built Vercel on top of AWS which charges him for bandwidth so he has to pass that cost onto his customers or he'll go bankrupt. Cloudflare, on the other hand, built our own network so bandwidth is largely free for us. The best judge of the 'quality' of bandwidth is how peered an organization is. Here's a list of the most peered networks in the world: https://t.co/WCHJ5siUid. Amazon is very good with about 600 networks it peers with, but Cloudflare has more than twice that number. That means that the default bandwidth you get for free from Cloudflare is actually more likely to be fast and reliable than the Amazon bandwidth you pay for from Vercel - **franpromano** (2025-09-04): https://x.com/franpromano/status/1963602273101119848 - I should have tag @irvinebroque - **franpromano** (2025-09-04): https://x.com/franpromano/status/1963601761752522920 - I vibe coded a cli tool to easily deploy workers to @CloudflareDev . prompts: .deploy or version release? .gets my env based on my branch . reupload env? bulk env file automatically .optional msg and tag (release) @dok2001 would be nice to have something similar out of the box. https://t.co/k82s4oJcTe - **elithrar** (2025-09-04): https://x.com/elithrar/status/1963590663485743521 - the whole “strategy vs execution” divide is stupid. it’s all execution. https://t.co/zBV6O5zEqk - **wesbos** (2025-09-04): https://x.com/wesbos/status/1963579103644110900 - Why did no one tell me about this?! I have discovered DIN rail and 3d printed wago clips https://t.co/JnL50rathd - **brian_scanlan** (2025-09-04): https://x.com/brian_scanlan/status/1963552743294967877 - Moving to PlanetScale added a 9 to our uptime. https://t.co/ZWRmazKbTC - **aidaniil** (2025-09-04): https://x.com/aidaniil/status/1963513380456501348 - my best performing tweets are all about cloudflare - **theweb3jess** (2025-09-04): https://x.com/theweb3jess/status/1963445052564209856 - There’s only 2 jobs at any pre-50 people startup: engineering and sales. Either you build that shit, or you go out and ask people to use your shit. Arguably, this is true up to 200 people. There’s no biz to op and product to manage. You yell across the fucking room to get shit going. Then patch up the next thing. Keep the ball rolling. Mentorship is organic. If you do well, people want to work with you. When they want to work with you, you get feedback. When you take in the feedback, you’ve grown. And that’s mentorship. And please god nobody say “I want to work in strategy” to my face. - **mjackson** (2025-09-04): https://x.com/mjackson/status/1963420889250517133 - I miss San Francisco - **yminsky** (2025-09-04): https://x.com/yminsky/status/1963403807158616147 - A new episode of Signals and Threads just dropped! This one is an interview with @ChrisLattner, talking about Mojo, a new-ish language for GPU programming that's aiming to be an alternative to the CUDA stack. https://t.co/jcWQ5O82Sl - **brian_lovin** (2025-09-03): https://x.com/brian_lovin/status/1963363803518963821 - Pretty awesome to do this kind of stuff directly in Notion databases https://t.co/ii95NPUDqg - **ankrgyl** (2025-09-03): https://x.com/ankrgyl/status/1963347774935814172 - As promised, some more extended thoughts https://t.co/5aK3gvLMQ5 The tl;dr is that software is changing dramatically, and one of the gifts of AI is that you can iterate at the speed of evals rather than A/B tests. Evals are non-trivial to build. figure them out, and you win. https://t.co/REkA3Ej80L - **dinasaur_404** (2025-09-03): https://x.com/dinasaur_404/status/1963287005078372829 - being a product manager kinda feels like being a detective... got all these clues leading me on different adventures 🕵️‍♀️ - **sebp** (2025-09-03): https://x.com/sebp/status/1963279545466749401 - I'm back to coding full time for work! I have re-joined @Cloudflare after I left there 9 years ago. I'll be working on their container product. Couldn't be happier to be working on problems of that scale again. - **aryasaatvik** (2025-09-03): https://x.com/aryasaatvik/status/1963275403759616339 - i am building something like an ai image studio but tailored for icons. the generated images are stored in R2 - i route the image requests to the deployed worker - with remote bindings i can pass the image url to FAL for image to image generations. as long as the route that proxies R2 resources doesn’t change i can use HMR with the rest of the codebase - validate cache control headers are working as expected D1 since i dont see it growing to need more than 10GB for the data - for SSR routes it helps to get a more realistic load time. i am using the same worker with SSR marketing pages and CSR app pages - **zeeg** (2025-09-03): https://x.com/zeeg/status/1963261251804447120 - Logs are here 🪵🪓 Just normal, plain, boring logs. Except we made them trace connected and they work beautifully with Seer and the rest of our stack. https://t.co/O1TgxWefBN - **aidaniil** (2025-09-03): https://x.com/aidaniil/status/1963157207698075896 - the urge to move the whole codebase on Cloudflare workers - **fidjissimo** (2025-09-03): https://x.com/fidjissimo/status/1963135975204467176 - So beyond happy to welcome Statsig and @vijayeraji to OpenAI! Vijaye is one of the best leaders I’ve had the chance of working with and I can’t believe I’m lucky enough to do it a second time. https://t.co/nux095Up4w - **thdxr** (2025-09-03): https://x.com/thdxr/status/1963071199094648935 - we finally built something that does SSR and it's the slowest feeling product we've ever built - **dok2001** (2025-09-03): https://x.com/dok2001/status/1963055168607682662 - Around 11 years ago, one of our competitors would tell customers we were lying about our scale and pointed to it being impossible with so few employees. https://t.co/4u2R5Y8kqs - **dok2001** (2025-09-02): https://x.com/dok2001/status/1963012099971621056 - If you've been in the same role with the same manager for 2+ years, start exploring new opportunities in or out of your company. Leaders: If you're not refactoring your teams every few years, you're missing chances to grow talent & innovate. - **ilyasu** (2025-09-02): https://x.com/ilyasu/status/1962995651916267804 - @WillManidis Mmm, we might draw different conclusions. 2015 FB had a very bright future. - **robin_j_brooks** (2025-09-02): https://x.com/robin_j_brooks/status/1962982244865220655 - The US Dollar has been essentially unchanged for 5 months now. This doesn't get talked about much, as it doesn't fit the consensus narrative, which is still very Dollar negative. But - as I have been saying for a long time - the Dollar fall is over... https://t.co/QzJAMGkmnI - **franpromano** (2025-09-02): https://x.com/franpromano/status/1962966773453471964 - @irvinebroque Brendan, any chance we have many envs in one worker like in pages? - **ilyasu** (2025-09-02): https://x.com/ilyasu/status/1962944318702026973 - Sort of amazing how much OpenAI leadership is mid 2010s FB leadership - **elithrar** (2025-09-02): https://x.com/elithrar/status/1962941203961266277 - Our dev docs team is hiring! We're looking for both an engineer to work on the underlying docs engine (as well as search, linting, tooling and perf) + a design engineering role to partner closely on docs experience, navigation, perf (always!) and UX. Links in replies. - **kevinweil** (2025-09-02): https://x.com/kevinweil/status/1962938974260904421 - 💥 I’m starting something new inside OpenAI! It’s called OpenAI for Science, and the goal is to build the next great scientific instrument: an AI-powered platform that accelerates scientific discovery. - **sduquej** (2025-09-02): https://x.com/sduquej/status/1962728605957661001 - @dok2001 @Cloudflare @dok2001 I bet @Fin_ai would do wonders for Clouflare very quickly. Great to hear you’re spending some time on this, I’ve had a couple of rough experiences in the past. - **thdxr** (2025-09-01): https://x.com/thdxr/status/1962630645655507369 - from day one opencode has been designed to be embeddable and composable we're officially launching the opencode sdk today here's an example launching a single instance and running a bunch of tasks in parallel - this is the same sdk the TUI uses so you can do anything it can https://t.co/YSOvgbAEAX - **BraydenWilmoth** (2025-09-01): https://x.com/BraydenWilmoth/status/1962571337656082443 - Durable Objects have 10 event hooks you can tap into out of the box with the Cloudflare Actors library. Sockets, storage, alarms – working with it is easy. Landing v1.0 soon, try it early and leave feedback. Features include... - `storage.sql` to execute SQL commands - `alarms.schedule` to schedule many alarms - hibernating websockets work out of the box - reference the instance `name` as a property - persist variable values through evictions - rpc with `Actor.get("identifier")` - configure with location placement - track all instances by name in a new instance - "destroy" an instance in a single command - no requirement to define a Worker - and more... - **developedbyed** (2025-08-31): https://x.com/developedbyed/status/1962274763637731759 - cloudflare hyperdrive cut my query time down by 110ms, what is this sorcery - **ccheever** (2025-08-31): https://x.com/ccheever/status/1962269126656680327 - Welcome back https://t.co/DkpkFYO020 - **Baconbrix** (2025-08-31): https://x.com/Baconbrix/status/1962268543829135843 - Moved to San Francisco! Ready to ship more than ever before 🚀 https://t.co/x5YMnmBASm - **michellechen** (2025-08-31): https://x.com/michellechen/status/1962217446728757400 - i found llama2 and i found windows xp https://t.co/8zR3yrNpZA - **nizzyabi** (2025-08-31): https://x.com/nizzyabi/status/1962161367466541072 - i hate what tech twitter has become a weird circle jerk of companies have formed who’s sole purpose is to impress others by posting features that will never come out their focus isn’t on building a company, but twitter hype and some retweets on their shitty company the unfortunate truth is that this path will only take you so far it’s limiting, it’s a bubble you were meant for more - **kevinakwok** (2025-08-31): https://x.com/kevinakwok/status/1962001319402881170 - I love that the industry arbitrage in old days used to be so simple Fashion: go to southern France and look at what people walking around were wearing Film: watch anime in japan and make live action sci fi film in US Anime: take mushrooms and do not tell the authorities - **abhiarya** (2025-08-31): https://x.com/abhiarya/status/1961976995493138616 - cloudflare workers are the best - **thdxr** (2025-08-30): https://x.com/thdxr/status/1961838055549018266 - two things people don't know 1. javascript is fast 2. new jersey is beautiful - **yagiznizipli** (2025-08-29): https://x.com/yagiznizipli/status/1961572630126817358 - I’m particularly excited about this. Soon we can just bind Rust methods to v8 directly in workerd codebase. This would allow us to write more Rust in @cloudflare workers! https://t.co/5AwSNxLfIm - **michellechen** (2025-08-29): https://x.com/michellechen/status/1961538095880556579 - last week i threw a little party for our label this week i shipped some new products and features for cf next week i go touch grass https://t.co/dwuyOInCtf - **darraghcurran** (2025-08-29): https://x.com/darraghcurran/status/1961536294162501707 - If you do plan to make it in person - I’ll be in town for this as will many our senior engineering and product leadership - come say hi! https://t.co/O3PxPZXgy7 - **destraynor** (2025-08-29): https://x.com/destraynor/status/1961528574860824764 - We're hosting an awesome event on Sep 11 (5:30pm) in San Francisco, with @cognition, @harvey, @Fin_ai (amongst others) It's focussed entirely on building frontier products in AI. It's both in-person and live-streamed, but if you're in the city, do come along! (🔗 in 🧵) - **ankrgyl** (2025-08-29): https://x.com/ankrgyl/status/1961525101658276196 - i continue to not be a fan of the direction model apis are headed. i think the beautiful thing about chat completions is that it does exactly one thing -- messages in and messages out -- and leaves the rest to the creativity of a developer. the next gen of APIs have lots of specific, built-in flags for specific features (web search, caching, reasoning) and lots of portability issues across providers. i genuinely do not believe the motivation is lock in (even though it seems suspicious), and instead i think they're just trying to get us access to new stuff that has an evolving surface area. but a swiss army knife api with 100 knobs is hopefully not the final solution. i'm hopeful that a new, appropriately extensible, pattern emerges. i think the best abstractions are extremely simple and flexible on the surface, with a lot of depth underneath. the current vintage of apis.... do not fulfill that premise. - **CloudflareDev** (2025-08-29): https://x.com/CloudflareDev/status/1961467854404161740 - And ICYMI.... 📣 You can now pipe raw WebRTC audio as PCM in Workers https://t.co/p89vYRX68K - **thdxr** (2025-08-29): https://x.com/thdxr/status/1961464128049975530 - like i've been saying - all of these ai programming products are built by people who hate programming https://t.co/dE68gT3lOy - **pk_iv** (2025-08-29): https://x.com/pk_iv/status/1961460335342445025 - good morning from the “axis of evil” https://t.co/HmuNI5Iwy8 https://t.co/o4Z9wJE5vF - **aninibread** (2025-08-29): https://x.com/aninibread/status/1961422392586023389 - One-click go swoooosh https://t.co/g4IZuuJpjq - **CloudflareDev** (2025-08-29): https://x.com/CloudflareDev/status/1961416104493854732 - Building Voice AI is complex. Today, we're launching tools to make real-time voice AI as easy to deploy as a static website, including: ✨ Realtime Agents ✨ WebSocket audio support for Workers ✨ Workers AI WebSocket support ✨ Deepgram on Workers AI - **martin_casado** (2025-08-29): https://x.com/martin_casado/status/1961221770633269581 - I honestly don't understand how something as obviously needed and banal as an identity layer could be controversial. While I may quibble with specifics this seems like a pretty well reasoned standard. - **martin_casado** (2025-08-28): https://x.com/martin_casado/status/1961215391524688153 - 🤔can someone explain this to me? I'm not involved in these companies (including Perplexity which Garry seems to be defending ... ) But this seems like an good step for the industry. Kudos for CF and BB for making an effort .. no? Maybe I'm missing something? https://t.co/q4DACOzDPr - **peytoncasper** (2025-08-28): https://x.com/peytoncasper/status/1961144765300486258 - some clarification is in order. the current method where you spoof your user agent, break obfuscated client side code and otherwise try to "act" like a human is not being taken away what didn't exist and is currently being defined is a reliable method for your ai agents to access the same internet you do while still preventing abuse from bad bots - **dok2001** (2025-08-28): https://x.com/dok2001/status/1961128144116879477 - Sorry @garrytan. The open internet needs standards to remain viable. Would you have same position for Reddit? This pushes forward a) a transparent, open standard where we help our customers verify the signatures and b) gives the website owner full control - they make the decisions on what to allow/disallow. Some people want agents to interact with their site - others want more control. We think it's up to the site owner to decide. We welcome conversation and debate for how to make it work for all sides. - **dok2001** (2025-08-28): https://x.com/dok2001/status/1961080822645035457 - Very excited to be working with @pk_iv @browserbase to help build a better Internet. https://t.co/rf3BHQErAF - **peytoncasper** (2025-08-28): https://x.com/peytoncasper/status/1961076394693124110 - it’s been funny talking to people as many didn’t believe this was possible. my mission is to make browserbase the most trusted browser infrastructure on the planet so that we can empower the next billion agent developers @Cloudflare ❤️ @browserbase https://t.co/BLaQskJK7Z https://t.co/snYUyMoDzN - **raman_idan** (2025-08-28): https://x.com/raman_idan/status/1961072826862993690 - @AnchorBrowser 🤝 @Cloudflare - The first-ever verified browser agent 🔥 Anchor and Cloudflare have partnered to deliver the win-win solution: a secure way for good bots to access websites for legitimate automation. ✅ No more flaky access for automation builders ✅ Traffic from verified bots that honor website policies, with optional pay-per-crawl This is one of many steps toward secure, reliable browser agents—and Anchor customers get it natively starting today 🔥 - **aninibread** (2025-08-28): https://x.com/aninibread/status/1961051767749099650 - Wanna yap to your website? 🗣️ Try adding conversational search to your website with AutoRAG and NLWeb, an open-source project by Microsoft. Now in experimental beta. 🌟 https://t.co/NBKVoWArew Have feedback? Tell us at nlweb@cloudflare.com - **zeeg** (2025-08-27): https://x.com/zeeg/status/1960821146343104562 - literal God mode error in our MCP server due to invalid schema on Sentry's downstream API I feel bad for anyone not using Sentry rn. We gotta make it easier to get the whole setup working correctly (our trace connectivity is what makes this special), but holy shit. https://t.co/j2VjUtUPql - **grinich** (2025-08-27): https://x.com/grinich/status/1960790763325071429 - MCP-UI is a novel project to add rich UI to MCP interactions This is super powerful and moving quickly in the MCP community Check out the demo from @liadyosef and @idosal1 @ WorkOS MCP Night a few weeks ago: https://t.co/G5LsWzAVGS - **svensauleau** (2025-08-27): https://x.com/svensauleau/status/1960700912055271703 - More models with less GPUs! https://t.co/pYrRFTEbls - **Cloudflare** (2025-08-27): https://x.com/Cloudflare/status/1960700100637860131 - Cloudflare built an internal platform called Omni. This platform uses lightweight isolation and memory over-commitment to run multiple AI models on a single GPU. https://t.co/LS6azDOQlp - **CloudflareDev** (2025-08-27): https://x.com/CloudflareDev/status/1960694071686533481 - Here’s how 👇https://t.co/CT1PRmmcfy - **michellechen** (2025-08-27): https://x.com/michellechen/status/1960693036570038765 - ✨ developer day for ai week! so many goodies: 1) new partner models from @leonardoai and @DeepgramAI now available on Workers AI 2) AI Gateway allows you to connect to providers without a key (load credits onto Cloudflare), secrets store integration for BYO Key, a request-response translation layer, better security with DLP scanning, and dynamic routing for more granular controls 3 & 4) technical deep dives on how we built a fast inference engine (infire) and model swapping techniques in-house (omni) - **CloudflareDev** (2025-08-27): https://x.com/CloudflareDev/status/1960691630249267542 - Learn more🔗 https://t.co/5D4XcJIjP5 - **CloudflareDev** (2025-08-27): https://x.com/CloudflareDev/status/1960690410042302466 - We just enhanced AI Gateway as a single pane of glass for: ✨Unified billing across model providers such as @anthropic, @googleaistudio, @groq, @openai ✨Dynamic routing of models based on optimized cost, performance, and user experience ✨Virtual keys for granular user-level access control and management ✨A client-side translation layer across providers and models. - **samgoodwin89** (2025-08-27): https://x.com/samgoodwin89/status/1960536370465923226 - @irvinebroque “backend” is the resource’s logical id (same across instances of the app), “my-app-backend-prod” is a physical name - it is the actual name of the resource in CloudFlare’s control plane for the “prod” stage of “my-app”. - **samlambert** (2025-08-27): https://x.com/samlambert/status/1960513282395267093 - trump puts a tariff on joins https://t.co/dw0KpkfJrg - **kentcdodds** (2025-08-26): https://x.com/kentcdodds/status/1960395376978223323 - This is a good article: https://t.co/ck050SvlaH - **zebassembly** (2025-08-26): https://x.com/zebassembly/status/1960394129944215993 - Go is such a horrible language, I am disgusted by the fact people willingly choose it to build systems. - **DKThomp** (2025-08-26): https://x.com/DKThomp/status/1960373685329801590 - There’s a huge wide lane for Normie Health Thought, which is that healthy diets are v powerful, exercise is genuinely amazing, building muscle as you get older is awesome …. and also, drugs are great, GLP1s are a miracle, mRNA is cool, vaccines work, and supplements are mostly bullshit (except for creatine and a few others) But for a variety of reasons, the online breakdown of health politics pits these two sides against each other, as if it makes any sense to have to choose between “lifting weights is good” and “the COVID vaccines worked” - **LukeW** (2025-08-26): https://x.com/LukeW/status/1960359313790374070 - people underestimate the degree to which “how applications work & are built” changes during technology platform shifts. https://t.co/fSDrc2K37N - **KennyJohnsonATX** (2025-08-26): https://x.com/KennyJohnsonATX/status/1960344320042573901 - Really proud of the team on this one. @Cloudflare MCP Server Portals are now live! ♾ Merge all your MCP config into one URL 🔐Enforce centralized policy for each server 🪵Centralize tool and prompt usage logging https://t.co/Wa6oze1Dqu - **G4brym** (2025-08-26): https://x.com/G4brym/status/1960342896747422124 - We at Cloudflare Lisbon team have been working on this for a few months, and we were eager to ship this to everyone! I really think this is solving most issues with MCP today, from server security, logging and discoverability https://t.co/ujm3o0gA8s - **awealthofcs** (2025-08-26): https://x.com/awealthofcs/status/1960335777159680095 - It still boggles my mind that we went from the entire housing industry being underwater to $22 trillion in home equity in a little over a decade https://t.co/vTXOQA4BCL - **__apf__** (2025-08-26): https://x.com/__apf__/status/1960174557823971400 - I recently quit a beloved engineering leadership role to become an IC at DeepMind. I want to become an expert again. the first time I was an expert, I worked my way up. now I'm starting over, but with more to prove & lose. it's terrifying and exhilarating at the same time - **matt_slotnick** (2025-08-25): https://x.com/matt_slotnick/status/1960097314862924121 - there seems to be this emerging take that the future of software is thin products that are hacking distribution and i sort of feel like that it's going to end up the opposite - **kevinakwok** (2025-08-25): https://x.com/kevinakwok/status/1960049694929031641 - Tomorrow @danwwang's book comes out. I cannot recommend it enough. Dan's writing and thinking is a delight. And he has been immersed for a very long time on understanding chinese industrialism, which is perhaps the most important topic of today https://t.co/GEPJCxNnTD - **zebassembly** (2025-08-24): https://x.com/zebassembly/status/1959766820711178614 - If you are on the Workers Paid plan you get 10M requests per month and then pay 30c per additional million requests. On top of that, we charge for CPU time (time spent not waiting for IO, average CPU time for request is usually <10ms) with 30K seconds included and then an additional 2c per thousand CPU seconds. - **matt_slotnick** (2025-08-24): https://x.com/matt_slotnick/status/1959761743628177762 - this is the perfect tweet. it's catnip for investors of every stage, personality, strategy, and level of prominence. everyone can find a way to take it personally and get really mad https://t.co/PI8IKmrbM4 - **paulg** (2025-08-24): https://x.com/paulg/status/1959714137988366715 - What's happening in America right now shows why it's good to have a world with multiple sovereign governments. Research on mRNA vaccines, for example, can continue in other countries even though it's being blocked in the US. - **realEstateTrent** (2025-08-24): https://x.com/realEstateTrent/status/1959663755589189653 - Best hamburger in America? I nominate Roam Burger (Bay Area) - **KennyJohnsonATX** (2025-08-24): https://x.com/KennyJohnsonATX/status/1959634070029160554 - @Cloudflare AI week is going to have some huge announcements. It was so fun building out some incredible new features dropping this week. 🚀 https://t.co/IK5iopJKmr - **romainhuet** (2025-08-23): https://x.com/romainhuet/status/1959317544772174186 - Working on a project this weekend? Try Codex CLI with GPT-5, included in your ChatGPT plan! ❯ npm i -g @openai/codex ❯ codex Pro tip! Use GPT-5 with high reasoning via this shortcut: ❯ codex -m gpt-5 -c model_reasoning_effort="high" Or type /model to pick interactively: https://t.co/824oLUKhsb - **steipete** (2025-08-23): https://x.com/steipete/status/1959249819475603714 - @bheemreddy181 What you want is https://t.co/RioBkoMyVm - **sarahtavel** (2025-08-22): https://x.com/sarahtavel/status/1958965063504863340 - I was definitely thinking in the consumer realm (and I agree, I think of OpenAI mostly as a consumer company). I suspect they'll dominate consumer for a while, but once the underlying infrastructure and ecosystem matures, we'll see more consumer entrants. (Writing a blog post on this...) - **sarahtavel** (2025-08-22): https://x.com/sarahtavel/status/1958962358543683898 - In the Web 1.0 era, a question that followed most consumer internet companies was "What if Google does this?" Of course, in most areas the question was silly (bc: focus), but not in all areas. e.g., RIP Mapquest, Hotmail, etc. Now the question is "What if OpenAI does this?" The only truly valuable "vertical" area that won against Google was travel. It was enough of a considered purchase and specialized enough that it demanded a different UI (of course, eventually Google got into it, but it required a large acquisition). OpenAI *needs* to go after some areas because they can't control what questions people ask, and they must feel a responsibility to handle those questions well. Health is a great example of this. And OpenAI has a huge advantage both bc they control the underlying model, and from a distribution perspective -- people are already there asking the questions. If you're a founder going after the same type of prompts, that is really tough to compete against. Certainly a lot easier to do that in areas where you want a separate space / identity / experience, or where ChatGPT's interface underwhelms for the use case. It will be interesting to see how this all plays out. In particular, Google struggled on new product execution and had to acquire companies that they could then superpower with its infrastructure and distribution. Will be interesting to see OpenAI flex their new products muscle over time.... - **scottlincicome** (2025-08-22): https://x.com/scottlincicome/status/1958947597478944929 - A terrible decision bad for almost everyone. Six quick reasons why: 1. Bad for Intel's long-term viability, as politics, not commercial considerations, increasingly drive its decisions. (SOEs are notoriously slow, bloated, & unproductive.) Foreign govts might also target it. https://t.co/XD7iGScOaO - **wesbos** (2025-08-22): https://x.com/wesbos/status/1958940109430301044 - I built a realtime LED grid and let the internet control it. Here is the stack: Cloudflare Durable objects Partykit React + Websockets ESP32 microcontroller 3D printing + lots of soldering https://t.co/vigTShu8L3 - **ScalingDevTools** (2025-08-22): https://x.com/ScalingDevTools/status/1958934765673087226 - New episode with the incredibly talented @ritakozlov, vp of developers & ai at @CloudflareDev 00:38 Containers release 04:10 Competing with the hyperscalers 09:20 Disruption 12:02 Cloudflare's values 14:52 Dev Twitter/X 17:04 Shipping fast at a public company 24:00 Agents 26:15 Feedback loops 29:32 Team structure at Cloudflare 31:51 Agents SDK (shipped first version in a weekend) 35:24 Acquiring/hiring all the best people - **fidjissimo** (2025-08-22): https://x.com/fidjissimo/status/1958927289892446429 - Pretty incredible https://t.co/nqtpx5aCNa - **rrnn** (2025-08-22): https://x.com/rrnn/status/1958899874151153990 - We’ve deployed MoQ in production! Long time coming and first step to broader adoption https://t.co/aFA9A7tOw1 - **Cloudflare** (2025-08-22): https://x.com/Cloudflare/status/1958878048566104354 - Media over QUIC (MoQ) is a new IETF standard that resolves this conflict, creating a single foundation for sub-second, interactive streaming at a global scale. https://t.co/XkeCcA2A4E - **mjackson** (2025-08-22): https://x.com/mjackson/status/1958689049440657653 - Ryan and I are barely writing any code ourselves for Remix 3. It will be the first web framework that is almost entirely written by AI. https://t.co/cSkuw1kRGk - **ajoshhoward** (2025-08-21): https://x.com/ajoshhoward/status/1958594714284040348 - `const stub = https://t.co/dkUg6VHs3c_DURABLE_OBJECT.getByName("foo")` You're welcome. - **camsinit** (2025-08-21): https://x.com/camsinit/status/1958546533223481656 - We just switched from Front to @intercom for exactly the reasons @eoghan lay out here w @lennysan Intercom did the hard work of becoming AI-native, and not only does it show, it’s felt https://t.co/cATxWzlJLc - **thomasgauvin** (2025-08-21): https://x.com/thomasgauvin/status/1958520413824127074 - Positive energy is infectious - **scottbelsky** (2025-08-21): https://x.com/scottbelsky/status/1958519123979178315 - true, any flex of self-awareness is a major draw to top early stage investors who know: bold journeys have a very messy middle, and only way to navigate the ups and downs is to acknowledge the risks up front and along the way with your team / board. https://t.co/lZgyePCZQG - **thomasgauvin** (2025-08-21): https://x.com/thomasgauvin/status/1958511786476400933 - I open sourced MCP Interceptor! 👇 It's actually a pretty cool usage of @CloudflareDev Durable Objects with Hibernating WebSockets, so if you've ever wanted to dip your toes into it, now's your chance 😎 https://t.co/lSWJpNBYro - **paulg** (2025-08-21): https://x.com/paulg/status/1958508895720399299 - Believe it or not, it's usually wise to walk investors through the risks involved in your startup. Investors know there's risk. If there wasn't, your valuation would be billions of dollars right now. And if you're vague about the risks you seem glib, or worse still, clueless. - **LegionHoops** (2025-08-21): https://x.com/LegionHoops/status/1958379306201325629 - PAIGE MF BUECKERS TONIGHT: 44 points 17-for-21 shooting (!!!!!!!) 4 rebounds 3 assists Wow. 🤯🤯🤯 https://t.co/NrFBPzZmms - **williamallen** (2025-08-21): https://x.com/williamallen/status/1958330997969494276 - I like our Browser Rendering product for many reasons, but a particularly important one is that we believe in being a good citizen of the web. There is no 'stealth mode' or trying to evade captchas - it always identifies as a bot. And, Browser Rendering's http requests are signed via Web Both Auth (the new open standard we've proposed for identifying crawlers & agents) to correctly identify itself since we first proposed it. - **kyliestew** (2025-08-21): https://x.com/kyliestew/status/1958328915090690175 - we just shipped a new templates gallery ✨ https://t.co/Q2gPo7JzEG - **yomnashousha** (2025-08-20): https://x.com/yomnashousha/status/1958270236563480782 - Calling all Cloudflare Pages users! 👋 Got an Astro static site? Would love to see it - drop links or DM me! Not using Astro? What's your framework of choice? - **kathyyliao** (2025-08-20): https://x.com/kathyyliao/status/1958237223956201684 - Automate structured data extraction with @CloudflareDev Browser Rendering /json endpoint -specify the expected response format -plug in your preferred LLM model with `custom_ai` -add automatic fallback models for resilience docs below https://t.co/ESFbORppWA - **williamallen** (2025-08-20): https://x.com/williamallen/status/1958219901291929939 - This is a big deal. So incredible to see the @C2PA_org standard continue to make traction in the world, driven by many, many dedicated people like @sherifhanna. https://t.co/cHuCJ3fheh - **patrickc** (2025-08-20): https://x.com/patrickc/status/1958212838604030196 - People sometimes say that the 12 most dangerous words in investing are "the 4 most dangerous words in investing are 'this time it's different'" but then I also wonder whether the 20 most dangerous words in investing are "the 12 most dangerous words in investing are "the 4 most dangerous words in investing are 'this time it's different'"". - **rgllm** (2025-08-20): https://x.com/rgllm/status/1958115598300328371 - @irvinebroque Sync worker settings with local wrangler. It could be easily done if you could export the wrangler file directly from the dashboard. - **0xJoker33** (2025-08-19): https://x.com/0xJoker33/status/1957938391703838750 - @irvinebroque when deploying a worker with queues, it wasn’t clear that the queue handler needs to be exported too from the message. using hono. after reading their docs i figured it out. - **threepointone** (2025-08-19): https://x.com/threepointone/status/1957928076815499648 - https://t.co/n6Z583kd5A https://t.co/nuqgaMe9vI - **threepointone** (2025-08-19): https://x.com/threepointone/status/1957832345081278787 - I'll be in SF on my birthday speaking at Sync Conf! Very excited, making a brand new talk for this one. https://t.co/Z4y6vyRxij - **ritakozlov** (2025-08-19): https://x.com/ritakozlov/status/1957801069968322770 - a feature i've been very excited about just landed in @cloudflare AI gateway: store your provider keys in our secret store to avoid having to pass them on each request💆‍♀️ https://t.co/m8aWiqpPc8 - **grinich** (2025-08-19): https://x.com/grinich/status/1957725882132398143 - @gabemonroy @Workday we have tried for years to get a dev environment to test our workday integration and we’ve been told time and time again that this is “impossible” please start there 🙏 - **danliu** (2025-08-18): https://x.com/danliu/status/1957533865607680482 - i think about this chart a lot. it's a great reminder of what product design is about. https://t.co/qn57lou48x - **thdxr** (2025-08-16): https://x.com/thdxr/status/1956558297638564116 - can't believe people are so negative on AI 15min of prompting while in bed and now have the perfect tool for this courtesy of v0 https://t.co/gVodgWfgpN https://t.co/RSgIDpkXpm - **williamallen** (2025-08-16): https://x.com/williamallen/status/1956514555120751041 - Some write for the superintelligence, others create for other humans and want that direct connection. Either is great - but you (the writer, the store owner, the content creator) should be in the driver's seat. That's why we, from the beginning, have given users the ability to choose for themselves. - **dok2001** (2025-08-16): https://x.com/dok2001/status/1956512295695032787 - Glad you are coming around to our POV. Better late than never. Different types of content have different purposes. Of course you want product documentation consumed by all types of AI, but if you are a creator that is monetizing content with ads probably not. https://t.co/RCqq8vS0SV - **saranormous** (2025-08-15): https://x.com/saranormous/status/1956444972141756544 - The era of superintelligence is here. Didn't predict the medical field would be first. Amazing work by team @EvidenceOpen in scoring a perfect 100% on the US Medical Licensing Exam https://t.co/9J8vOYGzHy - **swyx** (2025-08-15): https://x.com/swyx/status/1956439984854167727 - watch me realize in realtime that @gdb is deadly serious about character-level tokenization and it's already working well for @pdhsu, @SKonermann, and @patrickc's @ArcInstitute 's 40B model that is the "GPT-1 of DNA" with biological language models (instead of natural language), we may finally be able to understand the causes of genetic conditions like Ehlers-Danlos syndrome, scanning the billions of base pairs of individual human DNA in a massively scaled up way only AI can do. Superintelligent healthcare may be the one thing that excites me more than superintelligent coding agents. - **CloudflareDev** (2025-08-15): https://x.com/CloudflareDev/status/1956422344320364842 - ICYMI /pdf endpoint in Browser Rendering allows you to generate the exact pdf you want. Customize: 📎 paper size 📎 margins 📎 headers/footers 📎 background rendering For the full list of options, check out the API docs 👇 - **treibthedev** (2025-08-15): https://x.com/treibthedev/status/1956361126540697822 - Before MCPs and using Playwright MCP for testing were a real thing, @EastlondonDev was already doing that through his work at https://t.co/RAzgniAPlr That being said, my two cents about AI-first development is: follow Andrew!! - **shai_s_shwartz** (2025-08-14): https://x.com/shai_s_shwartz/status/1955968602978320727 - Are frontier AI models really capable of “PhD-level” reasoning? To answer this question, we introduce FormulaOne, a new reasoning benchmark of expert-level Dynamic Programming problems. We have curated a benchmark consisting of three tiers, in increasing complexity, which we call ‘shallow’, ‘deeper’, ‘deepest’. The results are remarkable: - On the ‘shallow’ tier, top models reach performance of 50%-70%, indicating that the models are familiar with the subject matter. - On ‘deeper’, Grok 4, Gemini-Pro, o3-Pro, Opus-4 all solve at most 1/100 problems. GPT-5 Pro is significantly better, but still solves only 4/100 problems. - On ‘deepest’, all models collapse to 0% success rate. 🧵 - **kentcdodds** (2025-08-14): https://x.com/kentcdodds/status/1955821579822338261 - Yes, I love seeing people learning that quality MCP severs require more care and thought https://t.co/ZWyljWXWA4 - **AnjneyMidha** (2025-08-13): https://x.com/AnjneyMidha/status/1955774094575853644 - I would watch a data center remake of The Office https://t.co/8IqaKcMXF0 - **bhalligan** (2025-08-13): https://x.com/bhalligan/status/1955748591538164008 - Founders are raising at crazy prices. I’m not sure they realize how unattractive it makes them to an acquirer… - **willahmed** (2025-08-13): https://x.com/willahmed/status/1955736935197978889 - I think whoop inadvertently killed alcohol consumption 😬 https://t.co/nHRijVqQxd - **matt_slotnick** (2025-08-13): https://x.com/matt_slotnick/status/1955722452459512133 - so this new CRM caters to this new business, of super efficient AI companies with lots of revenue and not a lot of people, tackling NEW markets - **matt_slotnick** (2025-08-13): https://x.com/matt_slotnick/status/1955721268592042497 - so when we talk about replacing that, you're talking about replacing process. it's not about the app. it's about process and good luck changing process - **matt_slotnick** (2025-08-13): https://x.com/matt_slotnick/status/1955720932045082681 - and, most businesses at scale also don't have one CRM. they have many sometimes dozens, across regions - **matt_slotnick** (2025-08-13): https://x.com/matt_slotnick/status/1955720855230849449 - the reality is that the modern CRM looks a lot more like byzantine business process management orchestrator than the contact management spreadsheet you're imagining - **korinne_dev** (2025-08-13): https://x.com/korinne_dev/status/1955699426657194391 - We're improving Environments for Workers, and especially how they work with our Git integration / the PR workflow. If anyone wants to provide feedback on our plans/designs, would love to talk! Here's my calendar: https://t.co/7CNUQUMnHF - **NotionDevs** (2025-08-13): https://x.com/NotionDevs/status/1955666043235692622 - Two insights from that first version: • Setup was too hard → most people never made it past step 1. • 1:1 API mapping made for clunky agent experiences (too many calls, too much token waste). So we went back to the drawing board. - **cherdougie** (2025-08-12): https://x.com/cherdougie/status/1955383195609993270 - "every company is launching a chain." and it's not even surprising anymore. stripe building tempo is like watching the final domino fall in what was always going to happen once payment companies realized they were essentially paying protection money to tether and circle for the privilege of using infra they could own themselves. i genuinely love stripe's approach to completely ignoring the crypto twitter discourse about decentralization and just focusing on what their customers actually want: cheaper, faster settlement with less friction. their fortune 500 clients don't give a singular shit about permissionless money or sticking it to central banks. they want predictable transaction costs and the ability to move money without waiting three days for ach transfers. stripe is basically offering them a private highway that still connects to the main roads when they need it, which is exactly what enterprise infra should be. all this goes to show distribution matters more than tech in the long run. ethereum has better tech than most of these corporate chains will ever have, but stripe already moves money for millions of businesses every day. they don't need to convince anyone to use their blockchain, they just need to make it incrementally better than what those customers are using now. liquidity follows users, developers follow liquidity, and suddenly you have an ecosystem that exists because it's useful rather than because it's ideologically pure. it's almost boring how practical this all is. - **joshk** (2025-08-12): https://x.com/joshk/status/1955371849346584727 - For the last decade, I have kept a Google Sheet with all my blood tests, medical records, etc. With Torch I have everything in one place. Ping @IlyaAbyzov to get on the Testflight today. https://t.co/HS6WGbXrYK - **IlyaAbyzov** (2025-08-12): https://x.com/IlyaAbyzov/status/1955363330358645226 - Working on Torch 🔥 Unified health record + LLM in one iOS app. Syncs records from hospitals, labs, Function, One Med, PDFs etc. Makes it simple & fast to get all your health data in one place + get LLM help making sense of it. Reply/RT for TestFlight invite. More below ⬇️ https://t.co/VBfoAMzGkJ - **BrankoMilan** (2025-08-12): https://x.com/BrankoMilan/status/1955127359494844458 - I do not know if anyone has written about this contradiction. New York City is the HQ of the globalized, commodified, penny-pinching capitalism. People go nuts if the stock market is up or down 1 percent. And yet many prices in ordinary life are simply negotiated, determined by personal relations, whim, or likes and dislikes. - **sejoker** (2025-08-11): https://x.com/sejoker/status/1955013817601003610 - One of the most thorough deep-dives we published in recent memory . Can’t be more proud by the team delivering the results in such a tight timeframe given the scale of the service. https://t.co/Jw6rmiPq2V - **zeeg** (2025-08-11): https://x.com/zeeg/status/1954980795850035384 - Straight forward authenticity is why I love Dax https://t.co/UdbdCJX9H9 - **dinasaur_404** (2025-08-11): https://x.com/dinasaur_404/status/1954917720220434433 - You can now deploy a platform on Cloudflare in just one-click! If you've struggled with setting up Workers for Platforms then this starter kit is perfect for you. Here's how it works 🧵 https://t.co/9aC18QmYdN - **BraydenWilmoth** (2025-08-11): https://x.com/BraydenWilmoth/status/1954913696465199503 - we're back to cookin' up those UI optimizations. @nandafyi is taking it to the next level. and yes "wedemboyz" is one of my databases. https://t.co/yZ004zNcFd - **zoink** (2025-08-10): https://x.com/zoink/status/1954633550978015515 - What would be your dream keyboard / set of input devices for @figma ? https://t.co/9CGNZtmrAw - **usgraphics** (2025-08-10): https://x.com/usgraphics/status/1954618122180428038 - Inexistence of task specific keyboards is a sign of an intellectually malnourished society. https://t.co/3FrFUWfTnk - **ankrgyl** (2025-08-09): https://x.com/ankrgyl/status/1954288217274110130 - I've been surprised and impressed by how all of the leading agents are essentially while loops with llm + tool calls. As a systems nerd, I've always wished that AI software (even pre-LLM) were modular. I love abstractions :) But the reality is, the bitter lesson of AI systems is that the more you can push decision making & control flow to the model, the better your system will get as models improve. Therefore, the simplest architecture wins. - **yusukebe** (2025-08-09): https://x.com/yusukebe/status/1954280232686678095 - Cloudflare ImagesがBindingで使えるようになったのアツいな。リサイズから回転、watermarkまでWorkersで完結する env.IMAGES.input(data).transform({ width:100 }) みたいにできる。なので、R2に画像を置いて、配信する時はリサイズする、さらにキャッシュもかけるってのができる - **yangcheng** (2025-08-09): https://x.com/yangcheng/status/1954120854872166508 - I think vibe coding requires even more engineering discipline. By definition, your code will be touched by multiple engineers (Claude Code), and the team has an incredibly high churn rate—each session is like a team member off-boarding and a new member onboarding. - **Baconbrix** (2025-08-08): https://x.com/Baconbrix/status/1953967090421158151 - This is huge. Expo CLI already exposes so much context via the CLI with everything from logs to errors to import stacks. Now the agent can automatically read it without needing an additional MCP layer. https://t.co/tlqe2yfgdb - **threepointone** (2025-08-08): https://x.com/threepointone/status/1953964774242627864 - RIP .dev.vars https://t.co/LMuA7lFhA6 - **southpolesteve** (2025-08-08): https://x.com/southpolesteve/status/1953960841499578863 - What should we ship next? https://t.co/q1XAuy3WIK - **CloudflareDev** (2025-08-08): https://x.com/CloudflareDev/status/1953960764467032141 - .env files for local dev environment variables and secrets now just work on Workers just like the --env-file flag in Node.js (or the dotenv NPM package before that) 🔗 below - **CloudflareDev** (2025-08-08): https://x.com/CloudflareDev/status/1953947267402281253 - If you liked the original: import { env } from "cloudflare:workers" you'll love the new sequel: import { waitUntil } "cloudflare:workers" https://t.co/n4SNyWK0Hv - **dok2001** (2025-08-08): https://x.com/dok2001/status/1953916502421238223 - We are not always perfect but over and over again @Cloudflare team responds to a crisis with transparency and follow through. And as an added bonus the additional resiliency comes with speed. Top = new first-party system Bottom = old third-party system https://t.co/ehQyraWNl5 https://t.co/FceSO8nsh3 - **stukennedydev** (2025-08-08): https://x.com/stukennedydev/status/1953897028720751096 - @threepointone Thought you may like these things I built for running specifically on Cloudflare and Durable Objects for Agentic workflows and memory. Fluxgraph - realtime workflow stuff https://t.co/Cm2IipH8Ja And a knowledge graph implementation that works nicely on D1 with visualisation. Also works well with fluxgraph https://t.co/NRJDPnu5ou - **brian_scanlan** (2025-08-08): https://x.com/brian_scanlan/status/1953872494537928765 - World-class uptime and performance is far from trivial to achieve when you're building using LLMs right now. You don't need all-the-nines of uptime if you're generating Ghibili pics or summarising a meeting, but if Fin can't answer because of LLM downtime, we lose money. - **CloudflareDev** (2025-08-08): https://x.com/CloudflareDev/status/1953856456878395437 - We just shipped a ton more detail to the Workers Builds page 😮‍💨 https://t.co/eXm4cdgfbT - **RozenMD** (2025-08-08): https://x.com/RozenMD/status/1953855861006160065 - so I recently realized we had builds data available for your Worker if you connected a git provider I've just shipped a change to the Deployments tab in Workers to display that data if available - each Worker version now shows you the branch name and commit that it came from https://t.co/Ht0TLrVKkf - **Cloudflare** (2025-08-08): https://x.com/Cloudflare/status/1953804662978588756 - Workers KV is Cloudflare's global key-value store. After the incident on June 12, we accelerated work to re-architect KV’s redundant storage backend, remove single points of failure, and make substantial improvements to KV’s p99 latency profile. https://t.co/lvIxTEqnEe - **zeeg** (2025-08-08): https://x.com/zeeg/status/1953692764283318649 - Should we just build a new Slack? Like, the true royal we. All of us. - **zebassembly** (2025-08-08): https://x.com/zebassembly/status/1953622397472440724 - Cooking up something interesting, here's a NodeJS Express + Handlebars + Postgres app running in workerd with 0 modification using an experimental bundler https://t.co/U6VsIrqmMc - **elithrar** (2025-08-08): https://x.com/elithrar/status/1953613313138733538 - the best PMs aren’t using AI to write PRDs. they are using it to do this. https://t.co/j5uOAfSAlV - **saranormous** (2025-08-07): https://x.com/saranormous/status/1953600723335487948 - .@eastdakota, cofounder CEO of @Cloudflare on the newest @NoPriorsPod today about how content creators should be paid by ai labs for filling in the “swiss cheese block” of human knowledge, the end of search, advice for web publishers, and different futures for the internet https://t.co/YCLMciuUZi - **nevikashah** (2025-08-07): https://x.com/nevikashah/status/1953583560923001096 - 🎨🧠 think in pictures like me? "Create a single mermaid diagram to render all subdiagrams for each of the proposed solutions. Add the person who proposed this solution in the title of each sub-diagram and a one line summary so it's easy to understand" thank me lataa 😘 https://t.co/alV9np3EwF - **bryan_johnson** (2025-08-07): https://x.com/bryan_johnson/status/1953540110374645892 - I paused Metformin. A longevity staple in my protocol for 5+ years. But its effects on my body were shocking. It built up my energy capacity, but limited my ability to use 90% of it. Here’s what we found. 🧵 - **CFchangelog** (2025-08-07): https://x.com/CFchangelog/status/1953531818017165450 - 🤯 Major Sandbox SDK update! Build AI agents, CI runners & more with live streaming, code interpreters (Python & JS!), Git support, & process control—all without infrastructure management. 🚀 Level up your Workers! https://t.co/ShuqAUKzqk - **elithrar** (2025-08-07): https://x.com/elithrar/status/1953514965404725337 - one of the most direct user facing teams in the Workers org! 👇 https://t.co/Na0PO5Ght0 - **thomasgauvin** (2025-08-07): https://x.com/thomasgauvin/status/1953443008159846869 - i built MCP Interceptor to debug and observe remote mcp servers it lets you see the actual json rpc messages that make up the model context protocol (without digging into logs) turns out mcp is just a set of 37 functions like "initialize", "tools/list", "tools/call" https://t.co/LzaBqhpuEb - **idosal1** (2025-08-07): https://x.com/idosal1/status/1953272652120158214 - Landed in SF! ✈️🌉 Excited to present MCP-UI at MCP Night 2.0 tomorrow alongside @liadyosef. Hit us up if you're in town! https://t.co/FdeCTzRmPs - **yagiznizipli** (2025-08-07): https://x.com/yagiznizipli/status/1953252596308951331 - Our library was a little bit (by 10 years) ahead of its time. No one understood its value… https://t.co/alEagQRdGZ https://t.co/8Re3r4mnut - **boristane** (2025-08-06): https://x.com/boristane/status/1953217858705469757 - optimism wins - **GaelBreton** (2025-08-06): https://x.com/GaelBreton/status/1953192139417866525 - @ivanhzhao This is a big deal. You guys could be the cursor of work. I really hope you guys consider making it an MCP client as well so we can connect all our data sources for work. - **ivanhzhao** (2025-08-06): https://x.com/ivanhzhao/status/1953180931260596387 - Looks familiar on the surface, but we completely rebuilt Notion AI. It will soon automate most things you do manually—dozens of pages a once, while you sleep. A small taste of what's to come to knowledge work🙂 https://t.co/PLkpbAHlkz - **corbtt** (2025-08-06): https://x.com/corbtt/status/1953171838382817625 - Announcing MCP•RL: teach your model how to use any MCP server automatically using reinforcement learning! Just connect any MCP server, and your model will start playing with it and (using RL) "learn from experience" how to use its tools most effectively! - **thomasgauvin** (2025-08-06): https://x.com/thomasgauvin/status/1953152726134829385 - If you're building on Workers, connecting to an external database, and aren't using Hyperdrive... why not & how can we make Hyperdrive better? lmk what you're building on Workers, I will read and respond to every comment - **dok2001** (2025-08-06): https://x.com/dok2001/status/1953128494386364687 - We have tracked this down. It was a bug in our UI that under certain rare circumstances showed a feature on when it wasn’t. It was NEVER turned on for ycombinator, but understand how @garrytan would have thought it was and been pissed. I’ve reached out to him to explain but haven’t heard back. We’ve resolved the bug and appreciate him flagging it. - **dcarter_js** (2025-08-06): https://x.com/dcarter_js/status/1953114935707193504 - ICYMI - Workers Builds disk space increase: 8 GB -> 20 GB. enjoy https://t.co/dGYqpwJLdY - **dinasaur_404** (2025-08-06): https://x.com/dinasaur_404/status/1953110625632014707 - New vibe coding platform from @Wix is built on top of Cloudflare for Platforms! 🤩 Uses Custom Hostnames for custom domain support + Workers for Platforms to deploy the sites! https://t.co/X05H4dqmiE - **zeeg** (2025-08-06): https://x.com/zeeg/status/1952935927556583461 - In addition to that, the burden crawlers put on every random property on the internet - to mostly serve the purpose of enriching their execs - is certainly not benefitting the producers of those properties. Look at _actual_ traffic to your personal blog. It’s insane. - **zeeg** (2025-08-06): https://x.com/zeeg/status/1952935923844620643 - There many more points on both sides that are easy to refute, but one absolute truth is that, generally speaking, web properties repurposing content rarely means the author of the content receives compensation. - **zeeg** (2025-08-06): https://x.com/zeeg/status/1952935918849179901 - A random bot crawling Sentry’s marketing pages a billion times a day has nothing to do with why ChatGPT recommends Sentry btw It’s the fact that we’re ever present in data that LLMs harvested to train on. It’s math. That’s it. This blocking debate is mostly nonsense imo. - **a85** (2025-08-05): https://x.com/a85/status/1952857267356959060 - Following Shopify's support for mcp-ui (https://t.co/oTIgfEg3Fz), Postman has also added support to both our agent mode and MCP client. This is a huge step towards making agents more powerful and interactive and I am excited about the potential of generative UIs! https://t.co/pg5AM5HrwG - **CloudflareDev** (2025-08-05): https://x.com/CloudflareDev/status/1952840735277105450 - Big news! 🚀 We've got @OpenAI newest open models on Workers AI! Access the new models with reasoning capabilities at @cf/openai/gpt-oss-120b and @cf/openai/gpt-oss-20b. ✨Support for the new Responses API ✨120B model is designed for complex tasks, deeper context understanding, and enhanced reasoning capabilities. ✨20B model is an efficient model for conversational AI and creative content gen. Read the blog & changelog 👇 - **michellechen** (2025-08-05): https://x.com/michellechen/status/1952838304191189326 - @ cf/openai/gpt-oss-120b @ cf/openai/gpt-oss-20b enjoy! - **ritakozlov** (2025-08-05): https://x.com/ritakozlov/status/1952832164631814640 - welcome two incredible new models: @openai's gpt-oss-20b and gpt-oss-120b to @cloudflare workers ai 🚀 https://t.co/eiaBtsx3Ha - **idosal1** (2025-08-05): https://x.com/idosal1/status/1952830331687473248 - INCREDIBLE to see MCP-UI widely adopted by a company like Shopify, just 3 months after I released it as an OSS community project. 🤯 Great to collaborate on it with the awesome @liadyosef. Amazing milestone! The future of the web is coming! Stay tuned. https://t.co/toaGoZKHkm - **tobi** (2025-08-05): https://x.com/tobi/status/1952800271257706676 - Agents will become a common way people shop. So today we are releasing 3 tools to make adding commerce to those agents trivial: - Checkout Kit: embed commerce widgets and checkout(!) directly into your agent and chat. This is already being used by Microsoft’s @Copilot. - Shopify Catalog: low latency global product search across millions of merchants - Universal Cart: shop from any store, anywhere, in one cart With these, you can build seamless embedded commerce experiences quickly, starting with all the best products by the best merchants in the world instead of having to sign them all up individually. No need to build a complex new checkout, or dealing with regulatory marketplace rules. Commerce just plugs in and feels seamless in conversations. We believe commerce should be everywhere people are, so we’re making that as easy as possible for the AI age. - **mottsmith** (2025-08-05): https://x.com/mottsmith/status/1952794848848957742 - Building permits are the most important leading indicator of growth in housing supply. The future in L.A. is very, very grim. L.A. Dept. of Building & Safety data dropped yesterday. https://t.co/7GEF2zE9B9 - **michellechen** (2025-08-05): https://x.com/michellechen/status/1952789248782024806 - On my way! https://t.co/ChsBMlSKSZ - **whoiskatrin** (2025-08-05): https://x.com/whoiskatrin/status/1952785418279149759 - 🧪 Major release for our Sandbox SDK! https://t.co/rnscW0icUy – live output streaming (see build logs or code execution in real-time) – code interpreter support for JS & TS &Python, with charts, tables & rich outputs (text, HTML, images, JSON) – preview your web service live using exposed ports (no infra needed) - full long running process control (start, list, kill, stream logs) - git Integration: clone repositories directly into sandboxes - session Management: maintain state across multiple operations - **whoiskatrin** (2025-08-05): https://x.com/whoiskatrin/status/1952785007963230289 - 🎈 we just shipped a pretty big agents SDK update: https://t.co/LpcNh6Q9Oh – MCP elicitation (yes to multi-step, interactive AI agents) – http-streamable transport (faster, more reliable than SSE) - auto transport selection (smart fallback between protocols) - improved error handling and reliable context updates - email integration improvements (now even better) more coming soon. go build cool stuff. - **DrJimFan** (2025-08-05): https://x.com/DrJimFan/status/1952747404379504855 - This is game engine 2.0. Some day, all the complexity of UE5 will be absorbed by a data-driven blob of attention weights. Those weights take as input game controller commands and directly animate a spacetime chunk of pixels. Agrim and I were close friends and coauthors back at Stanford Vision Lab. So great to see him at the frontier of such cool research! Congrats! - **ritakozlov** (2025-08-05): https://x.com/ritakozlov/status/1952745481953534213 - last week i asked @cloudflaredev PMs how they use AI day to day this was one of the best meetings i’ve been part of recently — there was so much creativity, way beyond the use cases you would think! here's what i learned: everyone assumes that AI for PMs is really in writing PRDs, but to my surprise, that was a use case not a single person had brought up in fact, towards the end of the meeting i asked: what’s something you tried using AI for, and found it wasn’t a good fit? writing PRDs was one of the responses don't offload your PRD (thinking) to the LLM the reason for that is that writing a PRD is a thinking process — it’s the process of defining the problem and trying to get at the root of what it is that you’re trying to solve for. you can’t outsource this part, and trying to shortcut it by asking AI to take a stab at writing a whole PRD only creates more work down the line while you shouldn’t have AI write the PRD for you, AI can be great at helping you write the PRD. here are few concrete examples in which PMs were using AI throughout the PRD writing process: - refining personas and what they need one way in which a PM mentioned they had used AI was to create very elaborate personas that they could then run the requirements through and see whether it satisfied them. if the personas have differing needs, you can then use AI to have them dialogue through their differences, and find solutions that address both (h/t @korinne_dev ) - defining states and edge cases; generating mermaid diagrams one of the hardest things to think through are all the states a user can end up in, and what should happen in each scenario. PMs used AI for: asking it to come up with all the edge cases that weren’t spelled out in the PRD that a user could hit thinking through solutions for the edge cases mapping out mermaid diagrams to help define user flows and journeys and visualize all the states one could end up in (h/t @yomnashousha) - competitive research a big part of PRDs is doing competitive research and understanding what state of the art looks like. LLM of choice + web search can expedite this process significantly, and provided the right prompt, can even give you pros / cons of different choices. prototyping this is the top response of what PMs are using AI for (!!) many PMs use tools like @lovabledev and @boltdotnew to prototype UIs — dropping in a screenshot of an existing UI and providing functionality in the prompt gets you a pretty long way. bringing prototypes to conversations with engineering makes it much easier to reason about behavior and expectations. (h/t @aninibread @yomnashousha ) some PMs have also started using Claude Code for similar prototypes with CLI (and even shipped a few minor features / changes!) (h/t @thomasgauvin ) building example apps with MCPs a big part of being a PM on the developer platform is writing many apps and examples to test out the developer experience AND provide better examples to users. previously, this could quickly become a major undertaking to get something up and running. today with AI you can ship examples relatively quickly to help refine the quality of the code, MCP servers (especially the cloudflare docs server) is great for making sure you’re using the latest information. another great hack was using gitmcp (go to claude, and just replace https://t.co/nF4BZgm9ZV with https://t.co/kF3e8EeQIP in the repo name) and add it as a tool. you can immediately use this as an MCP server that has direct access to a repo (h/t @dinasaur_404 ) summarizing looooong threads this was my favorite example of the day. one of the most stressful experiences as a PM is finding yourself in a 3 hour meeting block while engineering continues discussion in a chat thread. it can take hours to comb through the conversation, follow along, and understand the proposals. a very clever idea here was to use AI to not only summarize the conversation, but generate diagrams of each of the proposed solutions in the thread and map them to the name of the person who proposed them. so you can get a clear view of what’s on the table, what the differences between the proposals are, and jump into the conversation or make an informed decision. (h/t @nevikashah ) PS: come join our team this is an exciting time to be a PM, especially on dev tools! if the above sounds fun, we're hiring! - **CloudflareDev** (2025-08-04): https://x.com/CloudflareDev/status/1952514120550420826 - $ npm i agents https://t.co/c55Avs5DB8 - **simonw** (2025-08-04): https://x.com/simonw/status/1952513828123771128 - My updated post about ChatGPT agent and how it exposes its identity through HTTP headers - I had originally thought it also triggered crawls from Yandex and Bingbot but that turned out to be my own Cloudflare configuration, entirely unrelated to ChatGPT https://t.co/nfIxcusPiG - **yomnashousha** (2025-08-04): https://x.com/yomnashousha/status/1952459092699619421 - 👋 I'm a new PM on Cloudflare's developer platform :) I'd love to connect with people using: • Pages (static sites only or with functions/bindings) • Workers with builds set up What's your experience been like? DM me! - **southpolesteve** (2025-08-04): https://x.com/southpolesteve/status/1952422489847689383 - Claude Code with the prompt: "You are my personal assistant and chief of staff. Self organize using markdown files" Has replaced the entire bucket of what you would label productivity software. 95% of my input is talking to it via @superwhisper https://t.co/s1BSqjyVff - **elithrar** (2025-08-04): https://x.com/elithrar/status/1952408189304996333 - thinking about doing a @cloudflare Workers meetup on Wed Aug 20th in our SF office. looking for interested lightning talk (~7 mins!) participants: if you've built something on Workers, in the JS ecosystem, or on top of R2 and want to talk about it, drop me a DM. - **janwilmake** (2025-08-04): https://x.com/janwilmake/status/1952400034391007432 - Introducing: HTML Workers - HTML-First Full-Stack Serverless 🤔 Hot take: React is overkill for 90% of full-stack serverless apps. What if we could write full-stack apps like this? 👇🧵 https://t.co/fVYMtgwuJ0 - **ritakozlov** (2025-08-04): https://x.com/ritakozlov/status/1952370831197081951 - on friday i asked PMs on the @cloudflaredev team how they use AI and got some incredible responses (none of which were for PRDs btw) had a long post typed up in draft, but lost it 😭 should i rewrite? - **thdxr** (2025-08-04): https://x.com/thdxr/status/1952369170537910488 - opencode's scope is to make ai tooling completely open source all 100% MIT licensed no caveats - **legionsdev** (2025-08-03): https://x.com/legionsdev/status/1952109961497936369 - @rauchg @vercel after changing from cloudflare to vercel my build time went from 8mins to 1 min : ) - **colinhacks** (2025-08-02): https://x.com/colinhacks/status/1951742120659984463 - Making AI resources auto-discoverable in package.json 📦 https://t.co/CiBinFV7ov - **brandinpodziem2** (2025-08-02): https://x.com/brandinpodziem2/status/1951455415184793715 - Veronica Burton Most Improved Player of the Year🤝🏼 - **zeeg** (2025-08-02): https://x.com/zeeg/status/1951454080653750490 - It’s wild people still use mongodb - **PeteButtigieg** (2025-08-02): https://x.com/PeteButtigieg/status/1951438711859859813 - The President sees a bad jobs report, and responds by getting rid of the person in charge of the statistics. This is how America becomes a third-rate country. https://t.co/yFghlwSaTA - **yagiznizipli** (2025-08-01): https://x.com/yagiznizipli/status/1951420886470234138 - I’m coming for you @fastifyjs (and @matteocollina) - **MikeNomitch** (2025-08-01): https://x.com/MikeNomitch/status/1951403892736200755 - Express.js coming soon to @CloudflareDev https://t.co/GCZONbzFMq - **ritakozlov** (2025-08-01): https://x.com/ritakozlov/status/1951391734576943348 - best message on friday afternoon from @yagiznizipli https://t.co/ZSeaWckBwI - **yagiznizipli** (2025-08-01): https://x.com/yagiznizipli/status/1951388316420518177 - Say hello to my very old friend @UseExpressJS running on @CloudflareDev workers! https://t.co/M2FEFFMlG3 - **brian_lovin** (2025-08-01): https://x.com/brian_lovin/status/1951385907086659618 - Your goal as a designer should be to move as high up the fidelity gradient as possible without sacrificing speed. You'll know you're getting better when you can output ideas at higher fidelity with the same speed. - **dinasaur_404** (2025-08-01): https://x.com/dinasaur_404/status/1951364041676562589 - Workers for Platforms isn't just about Workers. You can use it to extend Cloudflare's entire developer platform to build a multi-tenant platform where every customer gets their own KV store, D1 database, R2 bucket, Vectorize, etc... with complete isolation built in 🧵 - **tylercowen** (2025-08-01): https://x.com/tylercowen/status/1951364006666776689 - Very, very bad to behave this way... https://t.co/SnjCTEwHh6 - **LJKawa** (2025-08-01): https://x.com/LJKawa/status/1951346459594703130 - No other way to slice it: this is incredibly fucked up. https://t.co/ISg9msm62t - **kyliestew** (2025-08-01): https://x.com/kyliestew/status/1951324603982512472 - Took a quick trip back to CO last weekend - Summitted Sniktau at sunrise + saw a goat - Climbed Meeker via the Iron Gates, looped back down Longs Peak’s standard route - Trail ran an unranked 13er over the Loveland Pass ski area Love trips like these 🩵 https://t.co/ezjNpjsN5Y - **kentcdodds** (2025-08-01): https://x.com/kentcdodds/status/1951322957357392308 - In today's issue of "Your OpenAPI Server is not an MCP Server"... Seriously, creating a good UX for MCP is no more a matter of "wrapping your API" than creating a good UX for a website is. More quality MCP servers please. https://t.co/6kCKvNxU9y - **ankrgyl** (2025-08-01): https://x.com/ankrgyl/status/1951316115126108535 - We are no longer charging per user. This applies to both free and pro plans. It is a privilege to be able to simplify our pricing. We've grown exponentially over the last year, which has made it obvious that the more you eval & log, the more value we can provide. Plain & simple. https://t.co/X3K6mNRpvL - **whoiskatrin** (2025-08-01): https://x.com/whoiskatrin/status/1951281308321173840 - Great Friday closing offers for the Agents team, some pretty fantastic people are joining this team - **ritakozlov** (2025-08-01): https://x.com/ritakozlov/status/1951263832342511915 - morning vibes: - rare cool summer day in nyc - freshly worked out - jimmy nardellos and other goodies secured from farmer’s market - smell of warm sesame loaf and bunch of basil wafting from my bag - tame impala in my headphones life is ok 😌✨ https://t.co/7rEbeiE7J6 - **bernhardsson** (2025-08-01): https://x.com/bernhardsson/status/1951098026967314434 - People should go public the way they want but when Modal goes public I’m going to push for some absurd esoteric auction system just for fun. https://t.co/VwVgUGH88M - **threepointone** (2025-07-31): https://x.com/threepointone/status/1950958351275209215 - “cloud development environments… find pmf through agents, not humans” are you getting it anon https://t.co/ln8vi7KeSB - **craigsdennis** (2025-07-31): https://x.com/craigsdennis/status/1950954320054337634 - 👣 On this week's Dry Run we had the delightful @ghostwriternr talking about what he's cooking up on @CloudflareDev platform. Spoiler, it's awesome. 🚀 We also speak about the many layers and uses of the word Vibe. Come hang out! https://t.co/jGkYrwqXfL - **Jack_W_Lindsey** (2025-07-31): https://x.com/Jack_W_Lindsey/status/1950952359485702238 - We found an unexpected mechanism that the model uses to sanity-check statements, where “concordance” and “discordance” heads check for agreement/disagreement between objects and properties. https://t.co/PnR817WXPX - **Jack_W_Lindsey** (2025-07-31): https://x.com/Jack_W_Lindsey/status/1950952355127820300 - Combining these methods, we can trace full computations through attention. For instance, we show how relevant query and key features conspire to direct “opposite heads” to the right token, and transform “small” features to “large” features. https://t.co/9hM5N6ciKD - **Jack_W_Lindsey** (2025-07-31): https://x.com/Jack_W_Lindsey/status/1950952357191352656 - We can also show the mechanism behind how models answer multiple choice questions — “I need an answer” features go hunting for “correct answer” features on the appropriate answer tokens! (Though we still don’t know where those come from) https://t.co/tGSNdhMIJk - **Jack_W_Lindsey** (2025-07-31): https://x.com/Jack_W_Lindsey/status/1950952346990862502 - Attention is all you need - but how does it work? In our new paper, we take a big step towards understanding it. We developed a way to integrate attention into our previous circuit-tracing framework (attribution graphs), and it's already turning up fascinating stuff! 🧵 - **celtics** (2025-07-31): https://x.com/celtics/status/1950922023665803470 - Payton’s handles go CRAZY 🤯 #NBAHandlesWeek https://t.co/iA5UFmBjoa - **CloudflareDev** (2025-07-30): https://x.com/CloudflareDev/status/1950678810774716485 - From the latest 2025 @stackoverflow developer survey, @Cloudflare is now the #1 cloud platform for developers who are "learning to code" and #2 cloud platform for learners using AI 🧡 We couldn't have achieved this without amazing developers like YOU! Your passion, projects, and community involvement made this possible. 🧡 - **mwylde** (2025-07-30): https://x.com/mwylde/status/1950592828717093304 - The stream processing team at @CloudflareDev is expanding! I'm hiring for senior+ folks to join my team to work on Arroyo, Cloudflare Pipelines, and our edge-native distributed log system. - **csparks1919** (2025-07-30): https://x.com/csparks1919/status/1950418712693002388 - @Gyurmatag @Cloudflare The Cloudflare MCP Servers do support auth with API keys. You can just use bearer auth with either a user or account owned token. You just need to make sure the scopes are correct based on the server, there's an example here using the observability server: https://t.co/D3ClZ2Qh3K - **rrnn** (2025-07-29): https://x.com/rrnn/status/1950292617960292439 - iceperf shows 10% better performance for the next version of Cloudflare TURN (tbd when we will release to production) https://t.co/5tadpZr9a5 - **EastlondonDev** (2025-07-29): https://x.com/EastlondonDev/status/1950266701028700213 - I like how AI coding lowers the barrier to making our own tools that much easier and nicer. Today I vibe-coded a vite plugin for cloudflare tunnels. Just set the hostname you want to use in the vite plugin and DNS and SSL will automatically be set up for you and it all routes to your laptop. Works with HMR and everything. Really great for stuff like webhooks and oauth redirects - **OpenAI** (2025-07-29): https://x.com/OpenAI/status/1950240348695072934 - As ChatGPT becomes a go-to tool for students, we’re committed to ensuring it fosters deeper understanding and learning. Introducing study mode in ChatGPT — a learning experience that helps you work through problems step-by-step instead of just getting an answer. https://t.co/B8VbRYJH6r - **Padday** (2025-07-29): https://x.com/Padday/status/1950228712827605048 - We have a very special @Fin_ai launch next week. Unlike anything we've done before, a first of its kind for us. It'll be very interesting to builders, founders, and technical leaders who want to stay at the forefront of AI. Please let your AI folks know, we'll be sharing a lot of technical detail about what we've built! Link in the first comment :) - **ritakozlov** (2025-07-29): https://x.com/ritakozlov/status/1950228600529310095 - huge congrats to @liadyosef and team! 🎉 @dok2001 and i used to talk about success for @cloudflaredev being the successful exit of a company built on the platform. crazy to see this come true! (and how much of it happened over a single weekend!) https://t.co/ZXp9M5wQkh - **destraynor** (2025-07-29): https://x.com/destraynor/status/1950206932171837950 - Here's an end to end demo of @Fin_ai. Jordan did this for a customer, it's unpolished, but if you're interested in Fin, AI CS, or just AI software in general, it's well worth a look. Timestamps 00:42 End user Fin UX 04:33 How you train Fin on content 05:40 How you train Fin on complex tasks (e.g. refunds) 06:40 How you test Fin 07:35 Deploy Fin across all channels (sms, email WhatsApp.) 07:53 Fin Voice (quick demo) 09:10 Analyse your conversations 09:45 Topic explorer (cool UI here) 10:20 Fin suggests edits to your help centre (my favourite feature) - **idosal1** (2025-07-29): https://x.com/idosal1/status/1950205674756976985 - We can finally share the great news!✨ (+@liadyosef) https://t.co/Q8of30ukO0 - **liadyosef** (2025-07-29): https://x.com/liadyosef/status/1950199499307004257 - Some exciting news 🚀 (+@idosal1) https://t.co/ZAJL218g23 - **BraydenWilmoth** (2025-07-29): https://x.com/BraydenWilmoth/status/1950193423438237780 - most people default think phone scrolling = bad. a habit we “want” to break. but i don’t think we attribute enough to how much more knowledge we pack in a scroll so long as you curate a good feed of content. we can learn faster than ever before because of this. - **akothari** (2025-07-29): https://x.com/akothari/status/1950176565834502170 - last year was the summer of releases. this year has been the summer of excellence. less sexy than shipping new features, but arguably more important than anything else we could work on. hope you’re noticing the vast improvements being made in @NotionHQ performance, reliability, quality, and all the little paper cuts. we've been hard at work! - **levelsio** (2025-07-29): https://x.com/levelsio/status/1950129790368858485 - I sincerely believe most Europeans think money does in fact grow on a money tree https://t.co/Qq7O0EHkVz - **willahmed** (2025-07-28): https://x.com/willahmed/status/1949875515403239804 - Americans now spend more on healthcare than groceries or housing. Something needs to change. credit @HealthcareAIGuy https://t.co/FQQvto8mIe - **artman** (2025-07-28): https://x.com/artman/status/1949873690541171165 - Code reviews haven't changed in ages. What would make them 10x better that we can address when building this for @linear? - **DhravyaShah** (2025-07-27): https://x.com/DhravyaShah/status/1949540884317716557 - Memory x Auth Just met the legendary, humble, smart @bekacru dude creates auth stuff for my entire company and wouldn't let me pay for his breakfast @supermemory is fully and proudly using better auth! https://t.co/tMHPoJPTgi - **rohindhar** (2025-07-27): https://x.com/rohindhar/status/1949490731120533639 - San Francisco apartment vacancy rate compared to the rest of the country and California https://t.co/DE9upsD7lJ - **zebassembly** (2025-07-27): https://x.com/zebassembly/status/1949482414595506573 - Nearly finished this Vite plugin that reimplements @nandafyi's cool debugger snapshot idea. By importing a module with a `type: "debugger-snapshots"` attribute you make it so any debugger statement under that part of the module graph will add a snapshot of the locals in scope alongside the location and time the snapshot was taken. Love this idea for illustrating how actual code works - **jeff_weinstein** (2025-07-25): https://x.com/jeff_weinstein/status/1948876386313154924 - Just wrapped a few days in Washington, D.C., meeting policymakers on @stripe AI and payments. 🇺🇸 - Every level of government—across parties—is taking AI seriously - Compute and manufacturing seen as critical - While many are optimistic on growth, job displacement needs solutions https://t.co/OAQWRjIU3N - **thomasgauvin** (2025-07-25): https://x.com/thomasgauvin/status/1948808833243251025 - beautiful friday in nyc https://t.co/1N4lNtGFMl - **rohindhar** (2025-07-25): https://x.com/rohindhar/status/1948786643806101732 - Does feel like the real estate market in Lake Tahoe is softening considerably all of a sudden https://t.co/TZ7NzL6euD - **dok2001** (2025-07-24): https://x.com/dok2001/status/1948476820489469970 - We are trying to figure out what is going one with @Starlink. @Cloudflare seeing irregularities in their BGP advertisements, and a complete loss of traffic https://t.co/vmw8yD0cgD https://t.co/x4mxiKCaik - **elithrar** (2025-07-24): https://x.com/elithrar/status/1948451254780526609 - why break the Web like this? we offered to run it almost a year ago but were turned away. https://t.co/49FK5wQBww - **troyhunt** (2025-07-24): https://x.com/troyhunt/status/1948174637441180034 - 180k requests per second is a pretty good go, thanks @Cloudflare! https://t.co/XPRS2DhGZC - **sundarpichai** (2025-07-23): https://x.com/sundarpichai/status/1948151726260867276 - Just got off the earnings call - great quarter and it was our (and my) 40th call as Alphabet. August will mark 10 years since we announced Alphabet. Been thinking about the incredible growth in our new businesses since then - Cloud, YouTube, Play, Subscriptions etc. To give a sense of progress, in 2015 all of Alphabet’s revenue added up to $75B. YouTube and Cloud alone ended 2024 at an annual run rate of $110B. - **garrytan** (2025-07-23): https://x.com/garrytan/status/1948121993401782672 - Real taste in products is not in the eye of other famous or noteworthy builders Taste is in being able to understand customers or users, model them and make choices that delight them and solve problems for them, and be right over and over again with incomplete information - **thdxr** (2025-07-23): https://x.com/thdxr/status/1948015397782909089 - grumpy engineers are often right and care about something important the reason they're grumpy is because no one else seems to care about it - it's why they burst out in rants the reason no one cares is because the grumpy engineer is missing the true source of the problem - **tconrad** (2025-07-23): https://x.com/tconrad/status/1948004589544047006 - I’ve been asked to be the next CEO of Sonos. Grateful for the perspectives and critiques that many of you have shared here these last 6 months while I was interim. So excited for the road ahead. - **darraghcurran** (2025-07-23): https://x.com/darraghcurran/status/1947964860689531225 - Jordan has been such a huge force in our rebirth as an AI first company - here are some insights he shared with some @BessemerVP portfolio companies about how we pulled that off. https://t.co/uTg7tuKgl3 - **p_lo** (2025-07-23): https://x.com/p_lo/status/1947849625710289078 - The Bay really 1 of 1, we dont even have summer at the same time as everyone else 🤣 - **fidjissimo** (2025-07-23): https://x.com/fidjissimo/status/1947843108013805956 - Think about the number of lives that will be saved once we have an AI-powered healthcare system. This isn’t theoretical anymore. “At Penda alone, AI Consult would avert diagnostic errors in 22k visits and treatment errors in 29k visits yearly.” https://t.co/GOxQnkK6Bf - **levie** (2025-07-23): https://x.com/levie/status/1947830617913364853 - One practice we’ve implemented in going AI-first at Box is every week someone demos their AI workflow to the whole company. Usually it’s something we never would’ve predicted. Good to balance central planning for tech and decentralization for use case adoption. - **kevinweil** (2025-07-23): https://x.com/kevinweil/status/1947830234147094723 - Doctors + AI > doctors. It makes intuitive sense but this is still a really big deal to see it validated irl. In the future you won't want to go to a doctor who isn't augmenting themselves with AI. https://t.co/csw7I92Eg7 - **kentcdodds** (2025-07-22): https://x.com/kentcdodds/status/1947798310934229136 - I'm building my own MCP client, and of course I'm doing it the hard way by using in-browser LLMs (https://t.co/UBrsD1KKng) and seeing how far I can get with those. Pretty remarkable it works at all honestly. I do not know how doing this well would be possible without XState. - **nxthompson** (2025-07-22): https://x.com/nxthompson/status/1947787211077824782 - The most interesting thing in tech: @Cloudflare has built a new tool that allows publishers to identify, track, and block any AI bots that are scraping their sites without permission. Here Gitesh Gohel and I explain the spreadsheet we've built to decide who to block --- and then we start blocking. It feels like an important moment where the balance of power is shifting back to creators and away from companies who have scraped with impunity and without any fair exchange of value. - **RedwoodJS** (2025-07-22): https://x.com/RedwoodJS/status/1947753152784019634 - 🤤Cloudflare just launched per-branch preview deployments! > wrangler versions upload --preview-alias staging - **jonas** (2025-07-22): https://x.com/jonas/status/1947747819424477464 - @dinasaur_404 Merge with normal workers UI - shouldn't be separate. Access to dynamic dispatch should just be a thing that you can toggle on for any worker and it should use the same workers UI as everything else - **CloudflareDev** (2025-07-22): https://x.com/CloudflareDev/status/1947725143888830930 - 🎉You can now run Browser Rendering locally with `npx wrangler dev`. This means rapid local testing, easy debugging, and no usage costs while you build. See more👇 https://t.co/2Ba7qySzNj - **elithrar** (2025-07-22): https://x.com/elithrar/status/1947447438794756308 - Have feedback for how we improve IAM & RBAC here @Cloudflare? 👇 https://t.co/BDRg3GSimw - **jhutchings0** (2025-07-21): https://x.com/jhutchings0/status/1947445137820553355 - Do you use @Cloudflare 's API? We want to hear from you as we're researching what to do next on our account API tokens. Let us know in the quick survey below! https://t.co/H1mSmqQFeq - **nikunj** (2025-07-21): https://x.com/nikunj/status/1947438681113169962 - Waymo spent 15 years getting from 90% to mostly autonomous. Companies are about to learn the same lesson.. https://t.co/XxFwzj856f - **PeterSaulitis** (2025-07-21): https://x.com/PeterSaulitis/status/1947378932065722799 - As an unabashed @webflow fan boy + @CloudflareDev marketer, today's Webflow Cloud announcement is monumental. Web designers can now add dynamism without janky JS / HTML implementations. All powered by Cloudflare. Thank you @bryantchou @leinwand + team. https://t.co/puVVSFUQqB - **ryenarussillo** (2025-07-21): https://x.com/ryenarussillo/status/1947303672062583209 - Over the years I haven’t loved some of the WNBA arguments for higher pay but now it’s pretty simple. If the Vegas Aces sell for $2m in 2021 and are supposedly worth $290-$310m now, why would you expect the players to receive just 9% of the revenue? The Valkyries expansion fee was $50m two years ago. Today expansion cost is $250m. In what other profession would someone as valuable as Caitlin Clark make only $78k? - **WalshyDev** (2025-07-20): https://x.com/WalshyDev/status/1946991389029634324 - Heyo, it's been a while since I've asked for feedback from people using @CloudflareDev What pain points do you have using our platform? This may be as someone just getting started, using our platform for a while or someone who has been using us for years. It's easy to forget pain points when you deal with them often, so lay it on me! How can we make you a more productive developer and make sure you can build the next big thing on Cloudflare? All feedback is good feedback :) - **FearedBuck** (2025-07-20): https://x.com/FearedBuck/status/1946732009159557483 - WNBA players showed up to All-Star warmups in “Pay Us What You Owe Us” shirts https://t.co/9hb15R52YU - **JerusalemDemsas** (2025-07-19): https://x.com/JerusalemDemsas/status/1946643965597372704 - What happened to Hochul? She's eons better in policy & messaging than just a year ago. https://t.co/wWZTSXE7DU - **martin_casado** (2025-07-19): https://x.com/martin_casado/status/1946623475223007376 - I've been vibe coding something to replace all the personal productivity software I use. It's roughly a daily notes / notion clone / personal CRM that uses github as a back end. What's remarkable is that this is such a common hobby project that whenever I want a feature I can normally one shot it. Already it supports: - encryption - entity tracking / linking - background research - calendar syncing - rich media support - alerting - automated personal pipeline creation - contact management - search It feels like barrier to vibe code this is lower than learning someone else's complex productivity platform. And a lot more fun. It may be the case that an entire class of apps gets squeezed between vibe coding and deep infra components (e.g. auth, olap, oltp, web crawling, etc.) - **mariokostelac** (2025-07-19): https://x.com/mariokostelac/status/1946568734086238256 - Eamon from my team has done some cool deep work on making llms go as fast as light! Almost. Link in the reply. - **kevinkern** (2025-07-19): https://x.com/kevinkern/status/1946529195368366102 - Codefetch supports now cloudflare workers. Means you can stream and load your repository in cf environment. Including markdown conversion. https://t.co/YmZxzGea4U - **grinich** (2025-07-19): https://x.com/grinich/status/1946391052489028082 - The new @SlackHQ API rate limits are absolutely insane: For accessing messages, you can now only make 1 request per minute, with a maximum of 15 messages 🤯 And yes this applies to ALL apps, not just public ones. We can't build internal tools on our OWN data. @Benioff this is wrong and hostile to your customers. Please reverse this immediately. If you are worried about abuse, just limit APIs for public apps. Companies must always have unlimited access to their own data in your products. - **dinasaur_404** (2025-07-18): https://x.com/dinasaur_404/status/1946244658931142887 - 🚀 Just built a "platform-as-a-service" starter kit using @Cloudflare's Workers for Platforms + Custom Hostnames 🧵 Deploy your own platform in minutes with: 🔀Subdomain routing 🔐Custom domains + TLS certs 📜Custom code deployment https://t.co/WgiBRgOvVY - **martindonadieu** (2025-07-18): https://x.com/martindonadieu/status/1946032163230515540 - Added Hyperdrive form @Cloudflare in front of @supabase for public endpoints. From 20% CPU to 2% I think I will be able to reduce the machine size from XL to medium. https://t.co/7A7XEJzjL0 - **thdxr** (2025-07-18): https://x.com/thdxr/status/1946013450854875498 - it's crazy there's this massive ecosystem of ios app intents - functions that take structured data and can execute an action on any app with things like auth solved and llms have not been able to do anything with them - **zebassembly** (2025-07-17): https://x.com/zebassembly/status/1945992120789069981 - What if your React components were multiplayer out of the box? No messing around with WebSockets, no client side state tracking, use your Actor as a component and you got multiplayer real-time UIs. Coming soon to a @Cloudflare react framework near you.... https://t.co/TGjhhxkYjZ - **threepointone** (2025-07-17): https://x.com/threepointone/status/1945869557693845928 - just did the math, and this month marks 20 years of my professional "career". I'm still just a programmer. still learning something new everyday. still feel like I'm catching up to everyone around me. and yeah, there's still so much more to _do_. here's to 20 more. - **garrytan** (2025-07-17): https://x.com/garrytan/status/1945669245318348822 - America will cease to be exceptional if the world’s best AI researchers can’t even be in the country https://t.co/WjVh3r4hJa - **DynamicWebPaige** (2025-07-17): https://x.com/DynamicWebPaige/status/1945640140598022640 - apropos of nothing: this is an appreciation post for @simonw, who tries out literally all the tools, and joyfully shares (for free!) what he learns in the hope that it will be useful to other humans grateful to have someone relentlessly building with the garage door open 💗🙌 - **annewoj23** (2025-07-17): https://x.com/annewoj23/status/1945636437392855522 - To the 23andMe Community, I am incredibly excited and humbled to share with you that TTAM Research Institute (TTAM), a new non—profit medical research organization that I founded and lead, has completed the acquisition of 23andMe.   I formed TTAM and pursued the acquisition of 23andMe because of my strong belief that everyone should continue to have the ability to learn about and benefit from their DNA, and that a non-profit structure is the best way to solidify our values and commitments to our customers, the scientific community and the world at large. Since its founding in 2006, 23andMe has stood for something revolutionary: that individuals should have a right to understand themselves at the most fundamental level — through their DNA.   Through their DNA they can understand their health risks , prevent disease and be true partners in participating in research that shapes the future of health and medicine. Understanding the code of life - our DNA - is one of the most exciting scientific missions of our lifetime.  Through our DNA, we can learn about our ancestry origins, prevent potential health issues, and help develop treatments that could benefit all of us.   Humans are 99.5% genetically identical, and we have DNA in common with everything alive today.  The pursuit to understand DNA will benefit all of us and all of life. That belief built one of the world’s most diverse and engaged genetic research communities. More than 15 million people have joined this journey, and over 80% have chosen to participate in research, contributing to over 275 peer-reviewed publications in cancer, cardiovascular disease, neurological disorders, and more.   These discoveries did not just stay academic - those discoveries were also translated into new genetic reports  for customers that helped them continue to learn about their own DNA.    Thank you to the 23andMe community that participated in research and contributed to discoveries that have had a meaningful impact for the world. I want to end on a personal note.   Over the last few years I unfortunately lost my father, my nephew and my sister.   These tragedies have sharpened my focus on what is most important to me and how I want to spend my time.   My passion is 23andMe.  The opportunity to give back to society with our research and help everyone benefit from learning about their genome with a healthier life, is a personal mission where I am dedicating my resources and my time.    We all have a disease or health condition that we care about .  By coming together as a single community, we are stronger and more powerful to make discoveries and to ultimately make a difference.   The future of healthcare belongs to all of us and it’s in our power to make a difference. I am honored to be back with 23andMe and look forward to sharing more with you in the weeks and years to come. With gratitude, Anne - **____hnsn** (2025-07-16): https://x.com/____hnsn/status/1945628396509434073 - The other night I couldn't sleep so I made a simple MCP server to let LLMs grab YouTube transcripts. Quickly get a summary, ask questions about it, etc. https://t.co/mblS14AiJ5 - **yagiznizipli** (2025-07-16): https://x.com/yagiznizipli/status/1945625028168495558 - Soon, http.createServer will be supported by @Cloudflare workers and you can use your fastify/express.js apps! https://t.co/PTj87qNuKH - **michellechen** (2025-07-16): https://x.com/michellechen/status/1945589829418238287 - ai researcher fantasy league backed by polymarket. who building this - **DhravyaShah** (2025-07-16): https://x.com/DhravyaShah/status/1945583048457949205 - @irvinebroque @supermemory fascinating... imaginining email workers at dhravya @ my.supermemory .ai I can send stuff to it or anyone else can and it stores it to the same, central store damn - **dok2001** (2025-07-16): https://x.com/dok2001/status/1945529733975843137 - Time to add a founder "acquihire" poison pill in all new software contracts - **CloudflareDev** (2025-07-16): https://x.com/CloudflareDev/status/1945521357460210053 - Dry Run: Ep 8: You've Got Mail https://t.co/JUdoggfQdX - **mwylde** (2025-07-16): https://x.com/mwylde/status/1945514667323052409 - R2 is becoming the obvious place to store your data lake, with low per-byte costs, no egress fees, and a managed Iceberg catalog. Amazing to see companies like @marimo_io realizing that and building first-class support for R2! https://t.co/LwWBWlilB7 - **marimo_io** (2025-07-16): https://x.com/marimo_io/status/1945509787745349743 - Use marimo notebooks to rapidly explore @Cloudflare data using Python and SQL. We've partnered with @CloudflareDev to integrate with Cloudflare authentication and to run marimo on Cloudflare Containers. Get started with examples for R2, workers AI, D1, and more. https://t.co/5RreIpwFQn https://t.co/GTfTR0WsaI - **PitchingNinja** (2025-07-16): https://x.com/PitchingNinja/status/1945491285860643109 - The Most Impressive thing I saw the entire All Star week? This 10 Year old girl has the Best Pitching Mechanics I've ever seen in a kid. 🤯 Literally. She's about 4'8" and topped out at 63 mph (without any warmup). https://t.co/caWIqZosa3 - **grinich** (2025-07-16): https://x.com/grinich/status/1945337448680108312 - acquired live @ radio city. sold out! what a night. congrats ben & david on achieving podcast rockstar status 🤘 https://t.co/4g0hchdByW - **dok2001** (2025-07-16): https://x.com/dok2001/status/1945330775466872941 - Blog post about 1.1.1.1 outage yesterday is live. This post goes into details about what led to the configuration change that caused the incident. It was not caused by a route leak. https://t.co/cPySEm1Kkv - **Cloudflare** (2025-07-16): https://x.com/Cloudflare/status/1945328334704918817 - On July 14th, 2025, Cloudflare made a change to our service topologies that caused an outage for 1.1.1.1 on the edge, resulting in downtime for 62 minutes for customers using the 1.1.1.1 public DNS Resolver as well as intermittent degradation of service for Gateway DNS: https://t.co/dyPaT11B25 - **kentcdodds** (2025-07-16): https://x.com/kentcdodds/status/1945301226981351899 - @irvinebroque Any chance we'll ever get eval/new Function support? It's required by elicitation in the MCP SDK - **TobinSouth** (2025-07-16): https://x.com/TobinSouth/status/1945285254513586337 - This is the correct response to visiting SF https://t.co/S7aCfKFbg9 - **davidlbowman** (2025-07-15): https://x.com/davidlbowman/status/1945271985107308840 - learning a lot about MCP at @Cloudflare from @sunglassesface at @ToolhouseAI https://t.co/YV682dLwIj - **nateberkopec** (2025-07-15): https://x.com/nateberkopec/status/1945229870256283927 - I'm sold. Agentic coding is the future of web application development. There is no going back. Close the editor. Open Claude. Your job is now to manage, review, corral and improve a relentless junior dev who is working on 6+ PRs in parallel. - **idosal1** (2025-07-15): https://x.com/idosal1/status/1945176527743623662 - Super stoked to speak at MCP Night 2.0 alongside @liadyosef! ✈️ We'll present the exciting future of agentic UI through demos of some of the hottest hosts and providers using MCP-UI! https://t.co/YLi4KQmUBM - **siddsax** (2025-07-15): https://x.com/siddsax/status/1945152924352500070 - Our @vercel bill before @Cloudflare, and after Cloudflare proxy. https://t.co/0XBfpKOcTW - **dok2001** (2025-07-15): https://x.com/dok2001/status/1945132870365851690 - Training and not new tools is still biggest gaps in AI utilization. - **grinich** (2025-07-15): https://x.com/grinich/status/1945123184207921297 - did you know MCP has a new UI layer? 🎨 join us at MCP Night 2.0 and hear from @idosal1 about their work on mcp-ui https://t.co/WSZicLf6hG https://t.co/RBDY5p9z9Y - **tobi** (2025-07-15): https://x.com/tobi/status/1945111518304981385 - I agree. We need people to create test runners that are more git diff aware, fail faster and provide prompt like context to aid an LLM in solving the problem. https://t.co/jFkVpLGMss - **jarredsumner** (2025-07-15): https://x.com/jarredsumner/status/1944949936447173042 - I think there’s a workflow here where you combine code coverage + faker-like data and you have a semi-automated fuzzer that fails if the LLM hardcodes fake data or the LLM fails to properly test code. Still need to experiment with this - **jarredsumner** (2025-07-15): https://x.com/jarredsumner/status/1944948478184186366 - `bun test` is going to get extremely good when using LLMs. Tests are the only way to verify code LLMs write works and are very important for their feedback loop Dogfooding this a ton right now - **clairevo** (2025-07-15): https://x.com/clairevo/status/1944921248615870598 - Real talk about running a startup: sometime you spend all day solving one user's problem, and it's 100% the right thing to do. - **JerusalemDemsas** (2025-07-14): https://x.com/JerusalemDemsas/status/1944889952736829528 - imagine creating a product for the sole purpose of maximizing the amount of calories from protein… then someone points out this is just fish… so you begin selling cod for THIRTY SEVEN DOLLARS A POUND https://t.co/pvMne1ra7l - **ZachLowe_NBA** (2025-07-14): https://x.com/ZachLowe_NBA/status/1944855683473850743 - New Zach Lowe Show! The LeBron/Lakers situation, OKC extensions, West outlook w/ @DanWoikeSports. Then @NekiasNBA and I dive deep on the key restricted free agents and the Pistons outlook: Spotify: https://t.co/muEF1dhydU YouTube: https://t.co/0yLx0fnZrA - **nateberkopec** (2025-07-14): https://x.com/nateberkopec/status/1944855203175600289 - Native structured logging PR proposal for Rails, extracted from Shopify. I'm not a huge fan of the gems available for this currently, so first class support in the framework itself (fiber-aware, phew!) would be a huge plus https://t.co/kIGrf6At5V - **CloudflareDev** (2025-07-14): https://x.com/CloudflareDev/status/1944827443724083529 - JavaScript Debug Terminals in VSCode, Cursor, Windsurf <> Cloudflare Workers breakpoint debugging now *just works* — no need to configure a launch.json file 🔗 below https://t.co/HNcLfqzxR9 - **CloudflareDev** (2025-07-14): https://x.com/CloudflareDev/status/1944827445443694735 - From VSCode, Cursor, or Windsurf: 1. open a debug terminal (cmd+shift+p and type "debug terminal") 2. run your local dev server ("wrangler dev" or "vite dev") 3. set a breakpoint in your code editor 4. make a request to your Worker 5. pause the world and debug with full access to the call stack https://t.co/cgC7ie8FYC - **berkie1** (2025-07-14): https://x.com/berkie1/status/1944792638575489056 - "I don’t care how much you love working people. They can’t afford a house because all the rules in your state make it prohibitive to build." @BarackObama https://t.co/zwUvzuUIli - **threepointone** (2025-07-14): https://x.com/threepointone/status/1944773465044611492 - you can now make a penpal buddy (Or customer service or a little personal assistant you keep sending invoices and stuff to whatever) super excited, the same agent can now host websites, receive phone calls and emails, you name it. give feedback! https://t.co/Gn1gtqpTwM - **CloudflareDev** (2025-07-14): https://x.com/CloudflareDev/status/1944744678227648856 - More Agents SDK updates just landed. $ npm i agents https://t.co/cow1u8Uhcw - **ghostwriternr** (2025-07-14): https://x.com/ghostwriternr/status/1944701214060413318 - New `agents` drop! 📨 Send or forward emails to your agents and have them take care of it! 🚥 `.queue()` tasks for agents to work on in the background 📡 Access your agent instance anywhere with `getCurrentAgent()` 🔍 Built-in observability for debugging and monitoring events https://t.co/ME78cxSTOa - **whoiskatrin** (2025-07-14): https://x.com/whoiskatrin/status/1944699819433320541 - 🎈 we just shipped a pretty big agents SDK update: - handle emails with onEmail() - queue tasks with .queue() - automatic context, so getCurrentAgent() just works docs and polish coming soon. go build cool stuff. https://t.co/OONvXWpEVg - **martin_casado** (2025-07-13): https://x.com/martin_casado/status/1944472724178674119 - There is an irony that Lina Khan’s activism has resulted in deals that are far worse for everyone except the people she was targeting. - **mattzcarey** (2025-07-13): https://x.com/mattzcarey/status/1944404334571123004 - I’d actually switch back to postman for this https://t.co/9oiZuIDEjK - **jonchu** (2025-07-13): https://x.com/jonchu/status/1944231394089349177 - I wonder if Sundar is making a mistake and this handicaps their ability to be effective leaders at Google. Hard to trust somebody and partner with them internally or externally when the founders so publicly stab people who trusted them with their livelihoods in the back. Senior folks at Google understand corporate politics and the ICs will go to other orgs vs working with a lead they can't trust and VPs will play games to protect themselves. But this is probably wishful thinking and the lower order bit - **kentcdodds** (2025-07-13): https://x.com/kentcdodds/status/1944221959715217639 - I repelled in a slot canyon today with my son and a few boys/parent from his gymnastics team. It was awesome. We did leprechaun canyon for those interested. Some spots were extremely tight! https://t.co/y3EtJOVn1D - **JustJake** (2025-07-13): https://x.com/JustJake/status/1944186601502126568 - I think the “shell-qui-hires” (Scale, Windsurf, etc) is one of the worst things to ever happen to startups The point of a startup is there’s assurance of binded fate “shell-qui-hires” undo that bindings As a founder, it’s your job to get everyone home. Full stop. - **ritakozlov** (2025-07-12): https://x.com/ritakozlov/status/1944155971200143670 - not hawaii, just staten island https://t.co/EGKYrDgpCR - **williamallen** (2025-07-12): https://x.com/williamallen/status/1944154320104304943 - I don’t think he’s ever told the story, but it’s worth telling. When we were selling @Behance to Adobe many years ago, @scottbelsky made a spreadsheet of every employee (32 of us at the time) and personally negotiated each persons title, salary and incentive structure, and made that part of the overall deal terms. I heard the phone calls where he went to bat for each individual. He not only didn’t have to do this, but it actually complicated some of the other factors in the deal. It changed the trajectory of so many people’s lives, including my own. 2 years later, 100% of that original team was still at Adobe. Even today, a dozen years later, many of the core members are still there, building. I was inspired by it then, and I'm inspired by it now. - **zeeg** (2025-07-12): https://x.com/zeeg/status/1944129309423415774 - Go touch water https://t.co/9esebTIsLw - **hearinladotcom** (2025-07-12): https://x.com/hearinladotcom/status/1944109825342026040 - Whatever you do, don’t show Stephen Miller this video of an adorable princess rapping to Three 6 Mafia in a Mexican restaurant. This is the United States in all its glory. https://t.co/GTQLH0gx6s - **zeeg** (2025-07-12): https://x.com/zeeg/status/1944073160602820644 - When these fake acquisitions happen they destroy what, imo, a core part of the valley is. When you take the best talent, the senior leadership, and you absorb them into another company and leave the others behind, in a lot of ways that is on par with just laying them off. - **zeeg** (2025-07-12): https://x.com/zeeg/status/1944073159109750888 - Sometimes I feel like this industry has forgotten what a lot of us are here for. "Windsurf employees did fine" People often join companies because of the people, the work they will get to do, the growth opportunities, AND the compensation. - **zebriez** (2025-07-12): https://x.com/zebriez/status/1944059826277167556 - If I were building out a People function today, I would have three sub-teams 1/ Team Fitness 2/ Recruiting and 3/ Company Vibes Team Fitness on individual performance/growth, top/bottom performer programming, manager/IC/leadership training, planning (gotta set clear expectations if you wanna drive individual performance), and traditional HR. Recruiting you get. But also employer brand like writing actually good job descriptions and helping teams talk about internal work as a beacon for talent. Company Vibes on internal comms, new hire onboarding, employee gatherings (hackathons, team/company offsites), swag - **rohindhar** (2025-07-12): https://x.com/rohindhar/status/1944053938329989397 - New downtown San Francisco apartment building has the fastest lease up in the city since 2013 50% leased in two months https://t.co/5lk1aYwtcb - **mattzcarey** (2025-07-12): https://x.com/mattzcarey/status/1943937910887133380 - mattzcarey .com now uses Astro. so painless to convert from nextjs. Feels really nice to add new blog posts. And hosted pretty much for free @CloudflareDev workers - **johnfreehayes** (2025-07-11): https://x.com/johnfreehayes/status/1943814976931852643 - @kevinakwok Long term damage to startup ecosystem here, investors get paid, first 10 employees get paid. Everyone else working for cash now. - **dok2001** (2025-07-11): https://x.com/dok2001/status/1943814380623475092 - Skeptical building teams with lots of egos works out as expected. Doesn’t work in sports or business. - **thdxr** (2025-07-11): https://x.com/thdxr/status/1943808486133182538 - kings used to lead the fucking battle charge and these little weasels are ditching their team for a w2 - **thdxr** (2025-07-11): https://x.com/thdxr/status/1943806219904594085 - this windsurf shit is such a bitch move as a leader you get a lot of upside but the deal is you go down with the ship - **dok2001** (2025-07-11): https://x.com/dok2001/status/1943804224652874031 - Is this the new norm? Founders get acquired and leave rest of team screwed over? https://t.co/ZvTiy3uYVr - **backpinelabs** (2025-07-11): https://x.com/backpinelabs/status/1943716747426595255 - Better Auth + Planetscale + Tanstack Start + Cloudflare is a seriously awesome stack. - **CSElmendorf** (2025-07-11): https://x.com/CSElmendorf/status/1943702086165930451 - The throughline? I don't believe people that who've got it good should be allowed to cluster w/their own kind & put up moats. I guess that's where I part ways w/ "Neighborhoods United SF." 4/4 - **CSElmendorf** (2025-07-11): https://x.com/CSElmendorf/status/1943702081111740683 - - Raise marginal income tax rate on high earners (or progressive consumption tax) - Fund universal health care - Fund universal rental assistance for poor people, w/ bonuses for moving to high-opportunity n'hoods /2 - **CSElmendorf** (2025-07-11): https://x.com/CSElmendorf/status/1943702083670081663 - - Establish & expand more national, state & city parks, wildlife refuges, etc. - Increase public investment in scientific & medical research - Increase immigration /3 - **CSElmendorf** (2025-07-11): https://x.com/CSElmendorf/status/1943702079329255462 - For the record, here are a few of my extreme right-wing positions: - Repeal Prop. 13 - Lower estate-tax exclusion from $15m to a few hundred thousand; increase marginal rate on large estates to 90+% - Tax carbon 1/4 https://t.co/zQneeTCyI2 - **samgoodwin89** (2025-07-11): https://x.com/samgoodwin89/status/1943695951282966874 - I re-wrote itty-aws to go all-in on @EffectTS_ It uses a Proxy and types-only, so the entire AWS SDK fits in 34KB. Every AWS API's errors modeled perfectly with TaggedError. https://t.co/hHV6Lbx7WN - **brian_scanlan** (2025-07-11): https://x.com/brian_scanlan/status/1943679716424130709 - It took me 10 years after leaving Amazon to finally have something equivalent to the retail orders graph at Intercom. It's fun when a single metric can change how an entire organisation works. https://t.co/gYNHzkvDzH - **itsaydrian** (2025-07-11): https://x.com/itsaydrian/status/1943668057454284991 - This is a @CloudflareDev appreciation post for being able to run everything locally while I debug this app. - **ryan_sherlock** (2025-07-11): https://x.com/ryan_sherlock/status/1943663604680520080 - I wrote a post on stop monitoring systems; start monitoring outcomes. TL;DR: Your latency and error rate dashboards can be all green, yet customers are having a bad time. Why? Because you’re not measuring what actually matters: customer outcomes. @intercom, with millions of conversations daily, reliability at its core is: can customers do their job? ⏭️ - **_mjmeyer** (2025-07-11): https://x.com/_mjmeyer/status/1943462244660187535 - holy edge-cannoli! 🤯 @CloudflareDev durable objects + @RedwoodJS server components = stateful magic at light-speed. my brain’s doing backflips... lfg!!! 🚀 https://t.co/ASVMfR0sKm - **destraynor** (2025-07-10): https://x.com/destraynor/status/1943419732516118811 - Spending a few hours on airplane wifi really convinces you that browsing the web via Claude/ChatGPT is a far far superior experience to web browsers for most queries. - **whoiskatrin** (2025-07-10): https://x.com/whoiskatrin/status/1943372906253128036 - i'm hiring on agents team! the crew is shipping like there is no tomorrow, you’d have a blast working together with @threepointone and @ghostwriternr. DM me if you want to join us - **clattner_llvm** (2025-07-10): https://x.com/clattner_llvm/status/1943364501417660466 - It was wonderful to get to spend time with the legendary @LisaSu today. She’s such an inspirational leader, it is no wonder that @AMD is on fire! 🔥 https://t.co/dlst8jsC0P - **Cloudflare** (2025-07-10): https://x.com/Cloudflare/status/1943342154547356155 - 🚀 New #ThisWeekInNET episode! Cloudflare’s Will Allen joins us to talk Pay-Per-Crawl and Content Independence Day (with a Matthew Prince cameo). How can content creators get paid when AI bots crawl their work? Big shift to opt-in. Watch here: https://t.co/T7rgK7YOAN #PayPerCrawl https://t.co/d3qlDFz4WX - **MikeNomitch** (2025-07-10): https://x.com/MikeNomitch/status/1943327980023148835 - Anybody waiting for inbound TCP on Cloudflare Workers? If so, would love to hear your use case @CloudflareDev - **G4brym** (2025-07-10): https://x.com/G4brym/status/1943327383227916745 - @ericclemmons @irvinebroque We already started publishing some endpoints that we feel conficent will not change after GA - **G4brym** (2025-07-10): https://x.com/G4brym/status/1943327079732224221 - @ericclemmons @irvinebroque the main issue preventing us from publishing the api's today, is that you still need to create and send us an api token for AutoRAG to read files from your r2 buckets 🫤 - **elithrar** (2025-07-09): https://x.com/elithrar/status/1942952062939025874 - personal favorite: AutoRAG just got WAY faster ⚡️ https://t.co/REck9N18Z8 - **samlambert** (2025-07-09): https://x.com/samlambert/status/1942796923196563503 - PlanetScale's 40 person engineering team has managed to build a more reliable database product than the rest of the industry. This includes multi-trillion dollar companies. - **samlambert** (2025-07-09): https://x.com/samlambert/status/1942780284858163272 - The project nobody has seen yet is Nova. Vitess for Postgres. Built from first principles, not using Vitess. We have spent 4 years sharding some of the world's largest databases. Those learnings will inform an architecture bespoke to Postgres. Only PlanetScale can do this. - **zebassembly** (2025-07-08): https://x.com/zebassembly/status/1942695106227609939 - Strongly considering moving @orangeframework away from react-router and leaning heavily into React Server Components. The primary feature Cloudflare can enable is distributed rendering with respect to data-locality, which better aligns with server components. - **tobi** (2025-07-08): https://x.com/tobi/status/1942628740909130167 - This feels like another step function increase in power for MCP. https://t.co/1vBAbelaZh - **liadyosef** (2025-07-08): https://x.com/liadyosef/status/1942619122015092772 - This is huge - we added remote-dom support to mcp-ui. This unlocks the possibility for clients to render UI from MCP servers *in their own style, look and feel*. Check it out. (thanks @tobi for commenting on this, and @_developit for some good feedback) - **korinne_dev** (2025-07-08): https://x.com/korinne_dev/status/1942619084165898374 - Remote bindings now work with Next.js! We're really excited about the ISR use-cases. Since the build step uses remote bindings to fetch from real data sources (KV, D1, etc), the initial cached page has fresh data from the start. https://t.co/CnGHFTNUU6 - **samgoodwin89** (2025-07-08): https://x.com/samgoodwin89/status/1942406622179103091 - New SQLite3 state store in Alchemy is nice! I like how I can just explore it in my IDE. https://t.co/JXPU1OZ3Oe - **warriors** (2025-07-07): https://x.com/warriors/status/1942350129782304884 - From: Coach Kerr To: Kevon Looney https://t.co/L4emDH3S6P - **CohenSite** (2025-07-07): https://x.com/CohenSite/status/1942336799772401676 - How is the built environment at the main entrance of one of the most famous and popular beaches in America literally metal shacks? https://t.co/HMX6BlEWPY - **zeeg** (2025-07-07): https://x.com/zeeg/status/1942303083343810676 - My current strong opinion is everyone who thinks you need different models or different tools to make codegen effective is just holding it wrong. Spend more time on prep. - **Nexuist** (2025-07-07): https://x.com/Nexuist/status/1942291985919778996 - One thing I wish I could beat into the heads of all the CS majors that follow me is that the Silicon Valley old guard stayed up late coding on weekends not so they could get rich enough to never code again but to buy more time staying up late coding on weekends in their 50s https://t.co/TAyripaCCe - **naumenko_roman** (2025-07-07): https://x.com/naumenko_roman/status/1942260644478341510 - Great good job with Alchemy IaC by @samgoodwin89 Users can escape IaC "jails". Example is Databricks. I'm helping somebody with devops practices. DB has a cli Go project with terraform provider, Python yaml generator on top. Alchemy gets it done fast, natively in Typescript https://t.co/UwboZi4Bqj - **kentcdodds** (2025-07-07): https://x.com/kentcdodds/status/1942231879932952994 - One of the critical bits for the true Jarvis-like AI Assistant for the masses future I envision is solving tool overloading through a "Google for MCP". People are working on this problem. Huge upside if it works. Here's how I envision that working. https://t.co/LPGctd1szt 📝 https://t.co/drwPqPj6o3 - **warriorsworld** (2025-07-06): https://x.com/warriorsworld/status/1941894905351262454 - Loon on Oracle https://t.co/bdfTqYHT93 - **levie** (2025-07-06): https://x.com/levie/status/1941892310075117580 - Huge opportunity for engineers, IT, or operations people to help companies implement AI Agents for enterprise workflows. If you’re on this site, you have a cheat code for what’s happening in AI that puts you ahead of the adoption rate of most orgs by a couple of years. - **MarkLevineNYC** (2025-07-06): https://x.com/MarkLevineNYC/status/1941655778005320070 - Congestion pricing by the #’s 6 months in: * 67k fewer drivers entering the zone daily * Delays in Holland tunnel down 65% * Traffic deaths down 32% so far in 2025 * $500m projected this year for MTA capital improvements * Broadway theater attendance up 12% * Subway ridership up 7% * Bus ridership up 12% * LIRR ridership up 8% So yes this program is a SUCCESS. - **grinich** (2025-07-05): https://x.com/grinich/status/1941605281873658282 - @irvinebroque @X Great question! The malicious link resolves to a redirection server that checks the user agent. If it’s “Twitterbot” (used for creating previews) it will redirect to Google Calendar. See screenshot. But when it’s your real browser, it will redirect to the attack URL. Sneaky! https://t.co/PUuHdxfJX3 - **threepointone** (2025-07-05): https://x.com/threepointone/status/1941593073080770956 - celebrating 17 years of meeting my wife the old fashioned way (being super drunk and hitting on her in a bar) things worked out https://t.co/UiLGAGgqgq - **grinich** (2025-07-05): https://x.com/grinich/status/1941545470595301446 - Just found a new account takeover attack on @X. Here's what to look out for: First you get a DM from someone claiming to be reporter that is interested to write an article about you. Flattering! Bluecheck account that is several years old, a bunch of tweets... seems legit right? They send a scheduling link to chat with their team. The link says Google Calendar, but it actually redirects to an OAuth consent page for a X app. This appears to be an official "Google Calendar" app, showing an icon and correct URL. But look closely at the permissions it's asking for: full account access. If you click "Authorize app" you have now given this random app full access to everything on your X account. If you do approve this, the app redirects to a real calendly page and let's you schedule an event. The invite eventually comes via a throwaway email address. I'm guessing most people never notice. If you later go look at your connected apps in X, this "Google Calendar" app seems legit. There's no way to see the name of the developer, any verified domain, or what API requests the app has done. It silently has full control of your account. Hopefully someone from @xdevelopers can get this fixed @elonmusk. Easy fix is to string match common names "Google" and also compare icons to popular apps to detect impersonation. The long-term solution is to build verification and trust for apps built on X. This will become even more important with AI systems and new features like calling and sending money. Please share this post so other folks don't get hit with the attack. And review/revoke any dubious apps you may have authorized. Stay safe! - **scottbelsky** (2025-07-05): https://x.com/scottbelsky/status/1941537194394694137 - I am surprised there is no single app that simply (1) collects all of your health data points - from whoop/oura data, blood tests, PDFs of lab tests you feed it, other scraped sources, AND (2) auto-generates a system prompt of sorts for any LLM when you ask any health question..? - **bryan_johnson** (2025-07-05): https://x.com/bryan_johnson/status/1941500440312733953 - 12 hrs after fireworks competing for most toxic air in the world purple areas = smoking 20 cigarettes ultrafine soot crosses blood–brain barrier toxins now entering water and food celebrating life, our customs bathing us in death https://t.co/JTy7rGsWlu https://t.co/9Av74eE8Zl - **__morse** (2025-07-04): https://x.com/__morse/status/1941220454288719891 - every new project exported with the React Export Framer plugin now creates a GitHub repository with CI setup to publish an example website using the Framer components This means that with one click you can get a full repo setup with your Framer components Try it! - **zebassembly** (2025-07-04): https://x.com/zebassembly/status/1941200658314809537 - @Zephraph @CloudflareDev How are you doing your API keys? The only way I was able to get wrangler to inline strings in the types was when putting it in `vars` which is not for secrets - **TobinSouth** (2025-07-03): https://x.com/TobinSouth/status/1940852844082745661 - Wanna chat about how we redesign identity for the agent decade ahead? Just co-chaired the first meeting of the AI Identity Management Group for the OpenID Foundation today. DM me if anyone wants to join the next one! - **thdxr** (2025-07-03): https://x.com/thdxr/status/1940814881969918150 - every ai tool is like "we spin up 7 subagents and make them fuck to the rhythm of a black horned tree cricket chirps and sample the results using an advanced algorithm (Math.random) and that's why we're #1 on this benchmark you never heard of" - **zeeg** (2025-07-03): https://x.com/zeeg/status/1940804244950622682 - At Sentry its a PMs job to make their product successful. There are not enough hours in the week to do that for 5 products, let alone more. IMO the era of PMs being a PRD factory has been dead in startups for a decade+, and LLMs aren't changing the narrative here. https://t.co/rX5KEs9IxS - **zenorocha** (2025-07-03): https://x.com/zenorocha/status/1940786229685244078 - We're starting the rollout of a massive performance improvement to @resend. • Before, the p95 was 1.28s • Now, the p95 is 229ms That's a 82% decrease in latency. This will affect our most critical API: the email sending endpoint (POST /emails). - **jamdotdev** (2025-07-03): https://x.com/jamdotdev/status/1940780619782672627 - Just launched! @Fin_ai + Jam Now @Intercom's AI agent can get bug reports from your customers https://t.co/iAWcDXemob - **carlrivera** (2025-07-03): https://x.com/carlrivera/status/1940757086259847173 - "Checkout" looks deceptively easy to implement: a few address forms and the ability to accept payments. Underneath it hides effectively all business logic a company wants to implement. As such, to build Checkout, you basically need to build all of Shopify AND get merchants to employ a second business system. This is especially painful for the larger brands who also happen to be the whale advertisers. We try to explain this to every partner, and in each case they have decided they must figure this out on their own. - **dok2001** (2025-07-03): https://x.com/dok2001/status/1940716508151509344 - Built in Bangalore! https://t.co/WUx4yMcvBE - **vaibhavshn** (2025-07-03): https://x.com/vaibhavshn/status/1940709693657342456 - Introducing CallsKit ✨ An open source toolkit to build realtime call experiences using @CloudflareDev Realtime. Has camera, mic, screen-share, chat, simulcast support. https://t.co/hHQDJjx0WS - **jonas** (2025-07-03): https://x.com/jonas/status/1940651441196077213 - In my mind LLM "agents" are just event-driven programs (or as @karpathy would say "Software 3.0" programs) 1) They can receive new events (e.g. a "user message") 2) Some events trigger a "computation" (i.e. an LLM request) 3) This produces additional events (e.g. an "assistant message") that other actors can subscribe to Under the hood the agents maintain an append-only immutable log and a reducer function that turns the event log into usable state - most importantly the input args for an LLM request (model options, system prompt, message history, tools) Modelling agents this way makes it easy to implement all manner of fun asynchronous interactions. Handoffs to other agents. Async tool calls (that complete in the far future). Scheduled events. Consuming webhooks / messages from third parties etc. And it's easy to do forking, snapshotting, time travel, debugging etc. And normal techniques (e.g. debouncing) from event driven programming (i.e. Javascript) are then easy to use, too. One thing I've recently been thinking is that you could even use a reactive querying mechanism like livestore or convex and give the agent a tool to query data. And then when the data changes the agent gets a new event. Are there any existing agent libraries that encourage this design pattern? - **dok2001** (2025-07-03): https://x.com/dok2001/status/1940611292446052641 - Ambitious, intelligent folks early in their careers: stop talking so much in meetings. Listen more. You were invited for a reason and should have nothing to prove. - **tbpn** (2025-07-03): https://x.com/tbpn/status/1940584851771736207 - We talked to @deedydas (Principal @MenloVentures and formerly founding team @glean) about startup talent and compensation. "It's not new that big companies throw out their wallets. But do $10M-a-year paychecks affect that? 100%." "At @glean we wanted people as missionary as you can be about, like B2B SaaS, but do $10 million-a-year paychecks affect that? 100%." "People are always on the spectrum for missionary and mercenary. And if the dollars go up, the spectrum shifts." - **jessfraz** (2025-07-03): https://x.com/jessfraz/status/1940581218598506836 - I wish I never saw this, I won't ever recover. The problem is not this guy. ITS YOUR FUCKING FILTER. - **jessfraz** (2025-07-03): https://x.com/jessfraz/status/1940578319713685642 - Absolutely no offense to the people who hired this person, but like, is this really your bar? I would just buzz past this. https://t.co/NOfpTU05wQ - **thdxr** (2025-07-02): https://x.com/thdxr/status/1940541957832327524 - anthropic is at risk of making a big mistake it's something we've seen too many times before imagine having the crazy goal of building a platform - something thousands of companies and products are built on top of you realize just building the platform isn't enough, so you start to build tools that make it easier to use the platform and demonstrate its capabilities these tools get their own names and identities and teams working on them and very quickly these teams forget they only exist to drive people onto the platform and then one day someone external makes a tool that accomplishes that goal and does it even better it should be a moment of success - this is the original dream, to see great things built for your platform but structurally these teams have long forgotten that so it's a moment of competition. in the worst cases they even try to squash it we've experienced this building SST and how some teams at AWS saw our work as competitive even as we were driving dollars to AWS and tapping into a market they could never reach there are exceptions - cloudflare has invested resources in helping us even though they have wrangler, somehow their teams are setup in a way to not see us as a threat but it's a real test - we'll soon be able to see if anthropic as an org is really aligned with becoming a platform or if they fall into this same trap - **williamallen** (2025-07-02): https://x.com/williamallen/status/1940471735872070140 - I came to @Cloudflare because I felt they were uniquely situated to build great products in a changing AI landscape. ✅ We brought @C2PA_org provenance to Cloudflare Images so the world can know what is real. ✅ We're proposing & implementing open identity standards for bots, crawlers & agents via Web Bot Auth. ✅ And now we've launched the ability to charge bots & agents for access at scale. More to come... https://t.co/0XBPf4MXzz - **BraydenWilmoth** (2025-07-01): https://x.com/BraydenWilmoth/status/1940195773976814048 - @mitsuhiko 🫣 soon (tm) I also have this package you can use that provides a UI to view your SQLite layer inside any of the instances you have. As easy as adding a class decorator above your DO class definition :) https://t.co/Sotul1J9SG - **mitsuhiko** (2025-07-01): https://x.com/mitsuhiko/status/1940195380970549611 - Until @BraydenWilmoth ships his teased explorer thing: Has anyone built anything fancy to access sqlite databases from durable objects for debugging? - **RedwoodJS** (2025-07-01): https://x.com/RedwoodJS/status/1940175591778427056 - Coming soon: Client side navigation (ViewTransitions), rendered on the server, and transported over websockets. https://t.co/f0IbwS8r2Y - **Stammy** (2025-07-01): https://x.com/Stammy/status/1940120674233983480 - 33 years old and the original Macintosh Human Interface Guidelines are still relevant https://t.co/1trY3R2bkJ - **boristane** (2025-07-01): https://x.com/boristane/status/1940101126084563241 - quality of life improvement your key worker metrics are right there alongside the list of all your workers https://t.co/UZcydypWZo - **garrytan** (2025-07-01): https://x.com/garrytan/status/1940082932435358177 - Keep skilled immigration legal https://t.co/FFOCNTkH16 - **eastdakota** (2025-07-01): https://x.com/eastdakota/status/1940082188097057199 - Watch this space as we track who is a good (and bad) AI bot: https://t.co/KliHprWBuO - **nandafyi** (2025-07-01): https://x.com/nandafyi/status/1940079163614052459 - went minimalist for Cloudflare's latest microsite https://t.co/C3FGqi19XU :) https://t.co/HjA0z9Lkql - **threepointone** (2025-07-01): https://x.com/threepointone/status/1940067725365453128 - what just happened https://t.co/38jpuOVGGW - **thdxr** (2025-07-01): https://x.com/thdxr/status/1940066216137093160 - gameover https://t.co/GeQ5mW8sZR - **igrigorik** (2025-07-01): https://x.com/igrigorik/status/1940059793584267463 - This is huge, and much needed. Bot traffic vastly outnumbers human requests. Bots are not bad, but identifying the good ones is really challenging. https://t.co/Jq566IKZIq - **eastdakota** (2025-07-01): https://x.com/eastdakota/status/1940050284446982479 - Proud of the Cloudflare team working to build a sustainable business model for the AI-driven web. Happy Content Independence Day! No AI-scraping without compensation. https://t.co/T4mKLFyTjr - **mindthegamepod** (2025-07-01): https://x.com/mindthegamepod/status/1940038064652554411 - That young OKC squad was ahead of their time. @KDTrey5 breaks down his old Thunder days and more on Part 1 of a new episode. Watch on our YouTube, @PrimeVideo or listen wherever you get your podcasts. https://t.co/5spQonrHD6 https://t.co/2s3yDxUTGV - **mindthegamepod** (2025-07-01): https://x.com/mindthegamepod/status/1940033999633854530 - There are few hoopers who love the game more than @KDTrey5. On part 1 of this special episode, he joins @KingJames & @SteveNash to talk about his development as a player and why he isn't surprised by OKC's current success. Watch on YouTube: https://t.co/5spQonrHD6 https://t.co/upkGlfCcN9 - **Cloudflare** (2025-07-01): https://x.com/Cloudflare/status/1940010017567994327 - Bots can start authenticating to Cloudflare using public key cryptography, preventing them from being spoofed and allowing origins to have confidence in their identity. https://t.co/nZEDZv5RwC - **ArmandDoma** (2025-07-01): https://x.com/ArmandDoma/status/1939903662903763157 - this is a profoundly stupid headline to describe a law that will massively reduce C02 emissions by building dense, infill housing, @nytimes! https://t.co/KV7K1Cg3Om - **glenmaddern** (2025-07-01): https://x.com/glenmaddern/status/1939901875752775707 - The original AI coding agent https://t.co/EDQQXUmGZo - **awong_SF** (2025-07-01): https://x.com/awong_SF/status/1939899119805636936 - In the 10 years I've lived in California, seeing the pro-housing movement become a force that it is, has been truly incredible to watch. https://t.co/QSEm5mOSyf - **tobyhardtospell** (2025-07-01): https://x.com/tobyhardtospell/status/1939898414810148963 - I can't believe that this takes effect *immediately.* What a baller move. We no longer live in a world where CEQA can block critically needed housing in California's cities. https://t.co/5E4n0pqidD - **GovPressOffice** (2025-07-01): https://x.com/GovPressOffice/status/1939893388113010771 - Turns out when California delivers the most consequential housing and infrastructure reform in modern state history, the whole country gets a push notification. Let’s build. ✅ https://t.co/ImbcHIVilu - **GavinNewsom** (2025-07-01): https://x.com/GavinNewsom/status/1939893132717695166 - California just delivered the most consequential housing and infrastructure reform in recent state history. It’s time for more housing across California — and it’s time to build it faster. https://t.co/mIM3TyUhUC - **JerusalemDemsas** (2025-07-01): https://x.com/JerusalemDemsas/status/1939888439517671904 - He literally shouts out @ezraklein lmao incredible https://t.co/knioNgR9Qo - **cornoisseur** (2025-07-01): https://x.com/cornoisseur/status/1939878169508479247 - @PEWilliams_ In the press conference, Governor Newsom said “Shout out to the NIMBYs, who will now be replaced by the YIMBYs” https://t.co/M5N21i7Cif - **PEWilliams_** (2025-07-01): https://x.com/PEWilliams_/status/1939875759318847863 - First national YIMBY push notification? https://t.co/ViDIeHeRwf - **DrNiravPandya** (2025-07-01): https://x.com/DrNiravPandya/status/1939857291135901725 - Kevon Looney overcame multiple severe hip injuries to play 290 straight games. He was a steadying presence for the team. He won 3 NBA rings. He did all the dirty work. He never complained. He shut down Sabonis. He rebounded like no other. We will miss him. #DubNation - **JerusalemDemsas** (2025-07-01): https://x.com/JerusalemDemsas/status/1939852420936343624 - I honestly can barely believe this happened. Total YIMBY victory in California today. After Newsom signs this bill; CEQA will no longer apply to *any* infill housing under 85’. No inclusionary zoning or labor provisions. https://t.co/N4L0VuBx4C - **CSElmendorf** (2025-07-01): https://x.com/CSElmendorf/status/1939850966599442899 - Incredible news. Two houses cleared, signature of @GavinNewsom to come later this evening. This is the most impressive case of gubernatorial leadership I have ever seen. https://t.co/MPEWhKhXsw - **ThompsonScribe** (2025-06-30): https://x.com/ThompsonScribe/status/1939826782125043960 - Looooooooon. End of an era. - **threepointone** (2025-06-30): https://x.com/threepointone/status/1939715464877220178 - here's the code to make a phone call to a cloudflare agent. twilio ⨉ openai ⨉ cloudflare. since it's running in code, you can add mcp servers, tools, cf bindings, whatever. front end is fully llm generated. have fun! https://t.co/JesQeHUe0l https://t.co/P0EqPuM4o0 - **sam_d_1995** (2025-06-30): https://x.com/sam_d_1995/status/1939683720492970067 - I am honestly amazed by how horrible this bill is. it simultaneously: 1) explodes the federal deficit 2)cuts taxes for millionaires 3) decimates Medicaid and rural hospitals 4) destroys American energy production it’s literally the worst bill of all time https://t.co/Q5KUtYvdpM - **mitsuhiko** (2025-06-30): https://x.com/mitsuhiko/status/1939632342345363549 - I did not publish this yet, because I'm not super familiar with how vite plugins should be published, but I would love feedback. It forwards console logs from the browser to the vite log. https://t.co/qUHIJi9os1 https://t.co/2XPWxorEfC - **whoiskatrin** (2025-06-30): https://x.com/whoiskatrin/status/1939608348074352695 - Oh, how happy I am this morning https://t.co/R11qlP98fO - **tannerlinsley** (2025-06-29): https://x.com/tannerlinsley/status/1939451404269289801 - 🚀 https://t.co/GBMjUXo9vg - **pbbakkum** (2025-06-29): https://x.com/pbbakkum/status/1939383162637070844 - A terrible and embarrassing thing has happened, which is that I’ve become MCP-pilled and now know it to be the future of everything - **johnarnold** (2025-06-29): https://x.com/johnarnold/status/1939347295545496005 - There’s no realistic scenario where the benefits (above the line) expire in 4 years or the pay fors (below line) are enacted in future. It's wrong to say this is a $3 tln tax cut for the wealthy paid in part by cutting benefits. In reality, it's just $6+ tln of unfunded tax cuts. https://t.co/ZS8aiXQd3J - **thdxr** (2025-06-29): https://x.com/thdxr/status/1939132317764886990 - you have to launch a lot of times here's the chart for claude code - their initial launch is bigger than most of us can do, we'd probably consider that a success but then there was a plateau and it took another launch to get it really going https://t.co/ouMDzX729s - **wunderacle** (2025-06-28): https://x.com/wunderacle/status/1939016621751988324 - I wanted (again) thank everyone from @crowdin and everyone that is a volunteer and has been contributing to Node's Website with translations and localization. You work is appreciated and you deserve a shoutout! - **garrytan** (2025-06-28): https://x.com/garrytan/status/1939012607362637894 - Mayor Daniel Lurie is fixing this city he deserves our support. Dmitry Yakoushkin is the deranged and mentally ill heckler who posted this video originally, and his activists want to destroy this city. Their behavior is a disgrace to SF and there's no place for it. https://t.co/peAWVUPMG7 - **thdxr** (2025-06-28): https://x.com/thdxr/status/1939005678170890504 - i think there's an (understandable) leap of if i make x more accessible, everyone in the world will do x but sometimes everyone tries out x and decides they'd still rather not do it also a lot of times x is made more accessible by drastically limiting what's possible we have every cooking gadget under the sun and food delivery is making more money than ever - **zeeg** (2025-06-28): https://x.com/zeeg/status/1939004383431401516 - That is, these CMS-style platforms adopt the technology to build more flexible no-code platforms. I dont believe we - in any rational horizon - see people with no software experience to launching a business that relies on the software functioning. - **JesseJenkins** (2025-06-28): https://x.com/JesseJenkins/status/1938974651453186516 - This is unbelievably bad. I am astonished that the Senate language got WORSE overnight than even the House version. This One Big Horrible Bill will raise energy costs, kill $100s of billions of new investment in energy & manufacturing, make our grid less reliable, increase pollution, and constrain our ability to compete with China for the future or AI. Total loser stuff. - **zebassembly** (2025-06-28): https://x.com/zebassembly/status/1938762042263118206 - Been thinking a lot about data locality and frameworks recently, I think with Cloudflare's edge and Worker's ability to split the application into multiple isolates across multiple data-centers we're in a really unique position to make this really simple. The edge can run your app by the user and by the database. - **LHSummers** (2025-06-27): https://x.com/LHSummers/status/1938746856000856470 - Rent control is the second-best way to destroy a city, after bombing. And, because of what it leads to in terms of under investment in repairing, maintaining, constructing new apartments, I think this is likely to exacerbate rather than improve issues around housing affordability in New York. This would be a gross policy error by @ZohranKMamdani. Watch tonight's Wall Street Week with @DavidWestin @BloombergTV https://t.co/DXrxYk1U1z - **otter401** (2025-06-27): https://x.com/otter401/status/1938740973263298859 - I've been very critical of @CAgovernor over the years for his lack of leadership on housing issues. And I need to give credit where credit is due. He stepped up big time getting AB130/AB609 to the 1 yard line, and hopefully over the goal line shortly. And @BuffyWicks is one of CA's best and bravest legislators in recent memory. - **cayimby** (2025-06-27): https://x.com/cayimby/status/1938699426656313840 - We are pleased to see that the latest budget language includes a clean CEQA exemption for market rate and mixed-income infill housing up to 85’ tall almost everywhere in the state; and we urge the legislature to pass it. - **cayimby** (2025-06-27): https://x.com/cayimby/status/1938699428518600834 - Huge thanks to @BuffyWicks, our partners in Labor, and the YIMBY movement for moving us closer to this win-win-win on housing affordability, labor, and the environment. - **yagiznizipli** (2025-06-27): https://x.com/yagiznizipli/status/1938659118677434435 - Some good news. We're becoming a first-time homeowners. Closing in less than a month! - **mrexits** (2025-06-27): https://x.com/mrexits/status/1938652269832266117 - Sam Altman on the illusion of hard work https://t.co/L7tZkuiOfA - **michellechen** (2025-06-27): https://x.com/michellechen/status/1938601650895163871 - https://t.co/jlCSaxPuDQ a library of all the ML papers i read + a chat so you can ask questions across different papers (with citations) not a novel idea, i just have the cutest domain for it - **signulll** (2025-06-27): https://x.com/signulll/status/1938600122855588311 - this is a high tier psyop from sam. he’s talking directly to the market. by floating a number like $100m: - he sets an anchor so absurd that anything meta offers now feels cheap. - he forces meta to either deny or match, both of which hurt them. - he gives his own employees a loyalty premium, they feel like they’re sitting on a $100m chip. - he blurs price discovery for everyone else. no recruiter knows the real going rate anymore. - & most importantly, he reframes the narrative from “openai is losing people” to “they’re so valuable it takes $100m to steal them” zero marginal cost, massive narrative leverage. & meta’s cto just confirmed it’s working by responding at all. textbook information asymmetry + narrative warfare. - **prisma** (2025-06-27): https://x.com/prisma/status/1938582096026751472 - 🦀 We are removing the Rust-based query engine from Prisma ORM and are replacing it with a more efficient "query compiler" written in TS. You can already try it for these DBs: ✅ PostgreSQL ✅ SQLite ✅ @Cloudflare D1 ✅ @PlanetScale ✅ MS SQL Server 👉 https://t.co/gW5x1ZtdEi https://t.co/mfOQoPKZw0 - **AlecStapp** (2025-06-27): https://x.com/AlecStapp/status/1938529615972053157 - Increasingly think this might be the most important chart in the world right now https://t.co/d3ELANKEhI https://t.co/f9xdHcWFXp - **marimo_io** (2025-06-27): https://x.com/marimo_io/status/1938472728044179948 - Our metrics show that people are hungry for better ways of working with data. Since our launch last year: ⭐️ 14000 GitHub stars 📈 334k downloads / month, growing 7% per week 📺 nearly 300k views on YouTube 🧑‍💻 150 contributors pushing code 💬 >2k members chatting in Discord https://t.co/5rLkGSqdpQ - **theo** (2025-06-27): https://x.com/theo/status/1938425955661582682 - Almost every agentic code tool spits out code that fails a basic Prettier format check. It amazes me that nobody else does this. https://t.co/tRxcLGgY1r - **thdxr** (2025-06-27): https://x.com/thdxr/status/1938424854249906263 - in opencode 0.1.150 it now autoformats your files whenever it makes a change there's out of the box support for prettier but you can add your own `file_edited` hooks in the opencode config encourage people to submit PRs with native support for other languages https://t.co/4xhzTaKS19 - **dok2001** (2025-06-27): https://x.com/dok2001/status/1938408599010660701 - fun fact: i have hired lots of people that way. https://t.co/R4CLcjdizC - **swyx** (2025-06-27): https://x.com/swyx/status/1938399666330341831 - icymi, openai also open sourced how their deep research prompt rewriter works + the full prompts today you can now build your own deep research agent - with full o3/o4-mini deep research quality - in 20LOC if you're a fan of Claude Research they ALSO dropped how to add multiagents with MCP support lmao - **chu_onthis** (2025-06-26): https://x.com/chu_onthis/status/1938378976168448449 - in more mcp news: the ts & python sdks have been fully updated to match the latest spec. try it out! yell at us if you run into issues! - **TysonTrautmann** (2025-06-26): https://x.com/TysonTrautmann/status/1938340413892661505 - Surprised at the +1s here. Auth is hard, bugs can be extinction events, enterprise customers demand roadmaps that eat engineering bandwidth at small startups. There's clearly a company size where it makes sense to build in-house, but outsource until then. https://t.co/BQUQ9yX1bC - **virattt** (2025-06-26): https://x.com/virattt/status/1938310481187401933 - @irvinebroque @harshil1712 @Cloudflare @AnthropicAI @findatasets No errors! Just blasted through the Worker usage limits for my plan. Upgrading today 🙂 - **GreenPlusAnE** (2025-06-26): https://x.com/GreenPlusAnE/status/1938306942067163513 - The US government taxes younger workers trying to start families, and gives their money to wealthier retirees. There is no moral or economic justification for this abomination. https://t.co/sJElZBqhdh https://t.co/wgJXtdPMaH - **ridd_design** (2025-06-26): https://x.com/ridd_design/status/1938287520908771475 - There's some magic happening in the @intercom design org right now 👀 I kid you not I took more notes from tomorrow's episode with @thoughtwax than any Dive Club convo ever... Think you're gonna like this one ✌️ https://t.co/7iNgxaHoRN - **CloudflareDev** (2025-06-26): https://x.com/CloudflareDev/status/1938275876254454074 - Run multiple Workers in local dev, one with Vite and another with Wrangler RPC, Service Bindings and Tail Workers now *just work* Lets you break apart apps into services that use different tools, and still run the whole thing locally 🔗 below https://t.co/GtuFP9cL5x - **mitsuhiko** (2025-06-26): https://x.com/mitsuhiko/status/1938148074016092165 - It's honestly impressive how well positioned Cloudflare is for the agentic future. This stuff just makes so much sense for where we're going. - **threepointone** (2025-06-26): https://x.com/threepointone/status/1938139217189618049 - gm https://t.co/JK5GN9mMxf - **mitchellh** (2025-06-26): https://x.com/mitchellh/status/1938066378185183279 - Re: infra, I've been out of the game too long. I don't think enough about it to have a good answer. But, I will say that the big cloud, TF, K8S, etc. era feels stuck in the past. Doesn't feel right. But I haven't thought critically enough about it to have a concrete idea there. https://t.co/G79XkFUCMg - **samgoodwin89** (2025-06-25): https://x.com/samgoodwin89/status/1938012506435506384 - MyContainer is Cloudflare's opinionated Durable Object for interacting with Containers. So good! https://t.co/zLOUE2fQZV - **samgoodwin89** (2025-06-25): https://x.com/samgoodwin89/status/1938003638070481339 - Another perfect example of @Cloudflare reducing a whole AWS service to `fetch`. Load balancing requests across containers with a simple rand() and fetch() 🤯 https://t.co/qK004SthVk https://t.co/qDtfVKM45T - **tomhacks** (2025-06-25): https://x.com/tomhacks/status/1938001702147461298 - Insights I learned this week: - Cloudflare Containers deploy any Dockerfile with ease - Wasmer runs "almost" anywhere, even in the Browser - Buildkit is wtf - OpenRouter is a simple facade to any model, not sure why @yacineMTB created his own thing +Good weather makes 𝕏 joyful - **wesbos** (2025-06-25): https://x.com/wesbos/status/1937994474367713306 - Is cloudflare sandboxes what will make me finally not hate docker? https://t.co/igJ69ULO6h - **MikeNomitch** (2025-06-25): https://x.com/MikeNomitch/status/1937987284785369138 - @morpiggg Note: We're still working on dialing this in though, so won't be perfect (especially if you spin up containers quickly), but baseline closeness will be better. - **MikeNomitch** (2025-06-25): https://x.com/MikeNomitch/status/1937984667778691569 - @morpiggg Following up - We identified the root cause on this and fix rolling out soon. Embarrassing bug on our end 🤦‍♂️ - **arthurcolle** (2025-06-25): https://x.com/arthurcolle/status/1937932909706678774 - @AnthropicAI It would be cool to make it easier to extract artifacts into runnable cloudflare containers for advanced dynamic dispatch - **kristianfreeman** (2025-06-25): https://x.com/kristianfreeman/status/1937902972668609009 - Sandboxes allow you to safely run arbitrary code on Cloudflare Containers. You can build a PR review bot that can safely checkout a git repo, install all the dependencies, and run tests. Thirteen lines of code. https://t.co/i1BD0VeJ46 https://t.co/smd7KqOMEy - **karpathy** (2025-06-25): https://x.com/karpathy/status/1937902205765607626 - +1 for "context engineering" over "prompt engineering". People associate prompts with short task descriptions you'd give an LLM in your day-to-day use. When in every industrial-strength LLM app, context engineering is the delicate art and science of filling the context window with just the right information for the next step. Science because doing this right involves task descriptions and explanations, few shot examples, RAG, related (possibly multimodal) data, tools, state and history, compacting... Too little or of the wrong form and the LLM doesn't have the right context for optimal performance. Too much or too irrelevant and the LLM costs might go up and performance might come down. Doing this well is highly non-trivial. And art because of the guiding intuition around LLM psychology of people spirits. On top of context engineering itself, an LLM app has to: - break up problems just right into control flows - pack the context windows just right - dispatch calls to LLMs of the right kind and capability - handle generation-verification UIUX flows - a lot more - guardrails, security, evals, parallelism, prefetching, ... So context engineering is just one small piece of an emerging thick layer of non-trivial software that coordinates individual LLM calls (and a lot more) into full LLM apps. The term "ChatGPT wrapper" is tired and really, really wrong. - **zebassembly** (2025-06-25): https://x.com/zebassembly/status/1937898908308373958 - It's taken years and lots of engineering, but with the launch of containers on Cloudflare we can finally run Workers on the developer platform. Workers have been something that Cloudflare has notoriously lacked, with containers I'm glad we're finally able to support them. https://t.co/lWzWFpT5FC - **charl_dot_dev** (2025-06-25): https://x.com/charl_dot_dev/status/1937888204075471286 - Apparently I made the Cloudflare mailer with one of my posts 🤭 One of my mates let me know https://t.co/8NQ1E513la - **southpolesteve** (2025-06-25): https://x.com/southpolesteve/status/1937886508066697598 - Can I buy cheap call booths from failed startups yet? Because I kind of want one at my house - **boristane** (2025-06-25): https://x.com/boristane/status/1937885915852284270 - run arbitrary AI-generated code on demand on Cloudflare Code sandboxes this is just the beginning https://t.co/5LF9xBBFqV - **_ashleypeacock** (2025-06-25): https://x.com/_ashleypeacock/status/1937884146375754122 - One of the use cases I talked about in my Cloudflare containers post was running code in a sandbox (e.g. a website to prompt an app needs to run "untrusted" code in a safe way). Looks like this is now easy thanks to a new npm package Cloudflare just dropped, aptly named sandbox! It has some nice capabilities such as being able to run commands inside the sandbox, you can make directories, checkout code from Git, as well as add/delete and move files. Seems a pretty nice wrapper, and I suspect there might be lots of simple packages to make common use cases easy with containers. - **whoiskatrin** (2025-06-25): https://x.com/whoiskatrin/status/1937883334526583275 - Cloudflare Sandboxes are live! https://t.co/sYtWNbPdWZ https://t.co/8ZIwq0x6za - **ghostwriternr** (2025-06-25): https://x.com/ghostwriternr/status/1937883000936525872 - More goodies for you - sandbox environments! Run AI generated code, close to your users, scaling infinitely. https://t.co/Oq7DieQKQ2 - **ritakozlov** (2025-06-25): https://x.com/ritakozlov/status/1937882759080321380 - we heard you liked containers, sandboxes let you spin up as many as you need to run all the untrusted LLM code personalized software is here! but needs a place to run securely... sandboxes are the beginning of us delivering on that vision (with support for workers coming soon!) https://t.co/nH3pyGZx4B - **elithrar** (2025-06-25): https://x.com/elithrar/status/1937882245337162070 - just launched Sandboxes today. you know, right after we launched Containers yesterday. go spin up a fleet of code agents orchestrated by our Agents SDK or Workflows and start shipping. https://t.co/quimKNFqxa - **CloudflareDev** (2025-06-25): https://x.com/CloudflareDev/status/1937880472404603176 - We're excited to introduce Cloudflare Sandboxes — the easiest way to run LLM generated code on demand. Sandboxes make it easy to spin up new instances of containers (and soon Workers — stay tuned!!) to make it easy to spin up tens, hundreds, or thousands of code-running agents https://t.co/p7Z3VIB1zL - **LHSummers** (2025-06-25): https://x.com/LHSummers/status/1937879125923880995 - I am profoundly alarmed about the future of the @DNC and the country by yesterday’s NYC anointment of a candidate who failed to disavow a "globalize the intifada" slogan and advocated Trotskyite economic policies. I fear it is evidence that our party is following the most problematic aspects of Britain's Labor Party. It didn’t work there, and it won’t work here. @Harvard and other universities' moral weakness after Oct 7 in condoning hateful, anti-Israeli, and even antisemitic, rhetoric and activity opened the Overton window in ways that contributed to yesterday’s outcome. I hope candidate @ZohranKMamdani , who showed great ability to learn and adapt during his campaign, will continue to evolve in ways that provide much needed reassurance to people committed to a free from prejudice, market economy as an American ideal. An important task for Mamdani now is to provide reassurance that his moral energy and formidable political skills can be productively directed for the City of New York at this difficult moment. - **dok2001** (2025-06-25): https://x.com/dok2001/status/1937878344173011350 - Well here you go... Cloudflare Sandboxes https://t.co/2RLYmG7flh https://t.co/EUkOGbKONY - **CloudflareDev** (2025-06-25): https://x.com/CloudflareDev/status/1937874920534601912 - sandboxes - **elithrar** (2025-06-25): https://x.com/elithrar/status/1937872942232428758 - we’re introducing a new way to build w/ Durable Objects: the actors library! it abstracts the boilerplate, adds a simpler SQL API, simplifies scheduling tasks & multi-DO architectures + you can drop to the underlying APIs whenever you need. dive in: $ npm i @cloudflare/actors https://t.co/qOdChJSbut - **KevinOConnor** (2025-06-25): https://x.com/KevinOConnor/status/1937869378806222870 - I didn’t know Aaron Wiggins had this in him but this is absolutely incredible https://t.co/6TCjMg9nYr - **DIYdoge** (2025-06-25): https://x.com/DIYdoge/status/1937868251293458466 - @irvinebroque Thank you for your response. I understand it's still in beta. If there's a way for me to view errors or debug the container so I can continue using it, I'd be happy to keep trying. - **ghostwriternr** (2025-06-25): https://x.com/ghostwriternr/status/1937865129468555397 - Cloudflare agents SDK 🤝 OpenAI agents SDK It's frankly beautiful how clean the abstractions are and how much work is taken away from you. With infrastructure and agent scaffolding solved, you can focus solely on what you need your agents to do. https://t.co/Ikv76k6fPB - **whoiskatrin** (2025-06-25): https://x.com/whoiskatrin/status/1937862644024680486 - you can now build AI agents using @OpenAI's Agents SDK and run them globally with @Cloudflare’s Agents SDK. pretty much everything you'd want for building real ai agents in one stack https://t.co/lLi6XaX61i - **BraydenWilmoth** (2025-06-25): https://x.com/BraydenWilmoth/status/1937862089332404402 - Today we're announcing the all new @cloudflare/actors library to help make developing with Durable Objects easier than ever before. Powerful feature defaults, improved interfaces, and a new default export handler. So what all can it do today? - Get current instance by name - Store all instances by name and last accessed - Route to a DO without a user-defined Worker - Define instance name within a DO - Execute SQL with template literals - Run SQL Migrations - Set multiple alarms by date, delay, or cron - Access an Actor by class (not by env.XY) - Destroy instances How to use it: `npm i @cloudflare/actors` Examples: ----------- Basic Worker ^ and yep. that's the whole index.ts file you need for the most basic Worker now. RPC into an Actor: ^ now you can just do a `.get` on your Actor class directly and pass in the instance name to interact with it via RPC. No more creating a stub and doing the things. Alarms: ^ schedule as many alarms as you want and define them with a Date, seconds delay (see above), or cron and pass in an array of parameters that map to the function you're calling into. Shout out to @threepointone for his work here. Storage: ^ define & run migrations, query with `this.sql` template literals. You can also see here how we log out `this.identifier` which logs out the user provided "name" value of the instance (not the ID itself). Also the second param in our `handler` allows us to define that we should in fact track all instances we create by name and store it in another instance altogether to keep track of. Shout out to @LambrosPetrou for his migrations work! Handler: As shown in all the examples above you can use our new `export default handler(...)` to export a Worker or an Actor. This handler does a lot of the magic of some of the features (such as tracking instances, holding identifiers and more). Best of all, these features can be used by your DurableObject classes as they are today without extending Actor... ...or.... you can just extend "Actor" instead of "DurableObject" and get all of the power this library unlocks automagically. ---- This is a BETA release to get user feedback. Intention of this library is to make working with various components of Durable Objects much easier, have less overhead, and add things here that may be more difficult or time consuming to add to the runtime to provide benefits to developers fast. Please provide feedback & share what types of features you would love to see in this library. Hope you all enjoy it! - **boristane** (2025-06-25): https://x.com/boristane/status/1937838071644405845 - I change my mind about things all the time one of the few constants in the world is change you're doing yourself a disservice if you don't change your mind when exposed to new ideas or information - **amiojinn** (2025-06-25): https://x.com/amiojinn/status/1937830473482551715 - @dok2001 The 2 "Configuration" should merge to one https://t.co/Pse1ZF1Yk6 - **eastdakota** (2025-06-25): https://x.com/eastdakota/status/1937828176056521011 - Cloudflare is the only DDoS vendor that could handle multiple simultaneous current-record DDoS attacks and not be crippled. Is your provider ready for what’s coming? https://t.co/jo7cE2nr6h - **threepointone** (2025-06-25): https://x.com/threepointone/status/1937815769439408328 - @zplesiv @jonas @appfactory @samgoodwin89 @benallfree hard problem, but something we're always looking at //cc @irvinebroque are you just looking for more memory atm? - **samgoodwin89** (2025-06-25): https://x.com/samgoodwin89/status/1937814360971518160 - @TobiM This doesn’t change what you can do at runtime. Alchemy is just modeling the configuration. - **samgoodwin89** (2025-06-25): https://x.com/samgoodwin89/status/1937812376939864303 - Much cleaner interface for Containers. Combines the DO Binding, Container config and Image build into 1. https://t.co/V2K0YwAvgW - **threepointone** (2025-06-25): https://x.com/threepointone/status/1937810612991508807 - don't host your website on containers, that's what workers are for https://t.co/VFHoiudFEh - **appfactory** (2025-06-25): https://x.com/appfactory/status/1937802908667367455 - @jonas @samgoodwin89 @benallfree We haven't done this yet, but we're going to add two methods: each container instance can be seen as a branch, so you can create a pull request on GitHub, or stash the code in R2 (which we might do automatically.) - **kylegalbraith** (2025-06-25): https://x.com/kylegalbraith/status/1937749823521972728 - If you’re looking at the price of Cloudflare Containers through the lens of it running 24/7, you might be holding it wrong. - **yusukebe** (2025-06-25): https://x.com/yusukebe/status/1937749171601310113 - Cloudflareブースにいるよ #AWSSummit https://t.co/Yn84AtbKY2 - **garrytan** (2025-06-25): https://x.com/garrytan/status/1937745257904669010 - Meanwhile in SF we are gonna build housing 🫡 https://t.co/fK90eO5H6j - **ramonashelburne** (2025-06-25): https://x.com/ramonashelburne/status/1937725377176371642 - Sam Presti takes every Thunder player to visit the memorial at some point in their career to understand what the city has been through. It’s really incredible to see it juxtaposed with photos of the parade. https://t.co/cGdMCTmjKN - **boristane** (2025-06-25): https://x.com/boristane/status/1937715672353767581 - the idea that investigating issues *during incidents* takes time may actually not be the true story the vast majority of incidents are the result of change identifying which change caused the failure is the real issue if you deploy/release often & in small batches, it's easy - **charliermarsh** (2025-06-25): https://x.com/charliermarsh/status/1937692702855893132 - https://t.co/pW09yQSSwE - **charliermarsh** (2025-06-25): https://x.com/charliermarsh/status/1937692675924185245 - Tonight's hack: deploy a uv-based project to Cloudflare Containers https://t.co/xNNjzZoBNx - **MikeNomitch** (2025-06-25): https://x.com/MikeNomitch/status/1937662381355581631 - @morpiggg Hey @morpiggg, something is off here, sorry. Unless you got very unlucky with prewarming placement (there's an element of randomness), we're doing something wrong on our end. - Team will dig into this - **mnafees** (2025-06-24): https://x.com/mnafees/status/1937636417758462369 - My entire feed is filled with news about Cloudflare Comtainers 👀 - **boristane** (2025-06-24): https://x.com/boristane/status/1937622865165123635 - this launched today btw https://t.co/mFyO6EtpZp - **DanielLurie** (2025-06-24): https://x.com/DanielLurie/status/1937611998868701204 - For generations, people in San Francisco have built the future. They've created businesses, families, and diverse and vibrant neighborhoods. But that promise—to build a life and belong to your community—is slipping away for too many. Too many parents are asking whether they can afford to raise their kids here. Too many young people are wondering if they’ll be able to stay in the city they call home. That’s why, today, I am proud to introduce our Family Zoning plan at the Board of Supervisors—so we can make San Francisco affordable now and for generations to come. This legislation will help us create the kind of neighborhoods we all want—safe, walkable communities, with small businesses on the corners and parks down the block. Here’s the truth: either we lead on housing, or we lose control. The state has given us a clear mandate, and if we don’t act, we risk losing funding and our ability to decide what gets built where. Our plan keeps that control in San Francisco, where it belongs. This plan is thoughtful but ambitious. It adds housing where it makes sense—on commercial corridors, on large lots, and near transit, schools, and jobs. And it does all of that while protecting tenants and maintaining the character of our neighborhoods. We want to build enough housing so kids growing up here will be able to raise their own families in San Francisco. We want businesses to stay and grow, confident their employees can afford to live here. We want firefighters, police officers, teachers, and nurses to live in the communities they serve. San Francisco’s future depends on the decisions we make right now. Let’s choose to grow with intention. Let’s choose to build a city that welcomes the next generation. I look forward to working with the Board of Supervisors and with communities across this city to get it done—and to keep San Francisco a place where families can put down roots and thrive. - **aidenybai** (2025-06-24): https://x.com/aidenybai/status/1937607276136992927 - if you're applying to a company, you should actually try their product - **kathyyliao** (2025-06-24): https://x.com/kathyyliao/status/1937599204291141918 - Browser rendering uses Cloudflare containers and now you can too! https://t.co/wu0HmzFWIE - **WalshyDev** (2025-06-24): https://x.com/WalshyDev/status/1937599007921934501 - SO EXCITED FOR THIS I've been seeing A TON of users asking for this and asking when we were releasing. The answer is now! Containers is in public beta y'all, come test and let us know what you think! https://t.co/gBTHBG8DTD - **RedwoodJS** (2025-06-24): https://x.com/RedwoodJS/status/1937597596115071030 - Framework fatigue is real. That’s why RedwoodSDK is the anti-framework. We give you browser APIs, code you can understand, and a deploy command that just works. - **IgorZIJ** (2025-06-24): https://x.com/IgorZIJ/status/1937590330204803343 - Just one example to add some color: building a new CI/CD platform is now trivial - mostly about UI. In fact I suspect there'll be a wave of startups exploiting this, reselling Cloudflare Containers compute at negative margin in exchange for rapid scale. Some of them will succeed in a big way (you can always move to cheaper compute all the way down to bare metal if need be). - **IgorZIJ** (2025-06-24): https://x.com/IgorZIJ/status/1937589128951902531 - This is a "holy grail" piece of tech. Many best practices of today are instantly obsolete. It'll take a while for the industry to fully digest it; enormous leverage for early adopters. You no longer need specialised compute services for long-running and / or full *nix workloads. You no longer need container hosting PaaS. Docker AND rapid start AND global AND long-running. Wow. Yes it's a bit pricey but the way of thinking of it from now on is: want it fast with minimal overhead? Use Cloudflare containers; willing to pay with operational overhead? Use _cheaper options_ like AWS / Azure. - **whoiskatrin** (2025-06-24): https://x.com/whoiskatrin/status/1937588132452663506 - @sargampoudel @irvinebroque @kuldeepsaini_23 oh you really need to stay tuned! it’s gonna be worth it - **dok2001** (2025-06-24): https://x.com/dok2001/status/1937587647393759602 - Nearly every acquisition at @Cloudflare has led to the teams building a second product with equal or usually substantially greater impact than the original. https://t.co/7xkXyPo0t6 - **skeptic_fx** (2025-06-24): https://x.com/skeptic_fx/status/1937585601936785720 - And the container platform itself is built entirely on top of Cloudflare Workers + Durable Objects 🚀 https://t.co/r8GylrGHpC - **samgoodwin89** (2025-06-24): https://x.com/samgoodwin89/status/1937584646604357639 - @MikeNomitch_CF https://t.co/1BUsUfUYAB - **EastlondonDev** (2025-06-24): https://x.com/EastlondonDev/status/1937584062073319732 - @nickbalestra @samgoodwin89 All of Cloudflare from Workers to Containers is about using lots of small instances effectively. Basically Zerg tactics Cloudflare: Zerg Google Cloud: Protoss AWS: Terrans - **dok2001** (2025-06-24): https://x.com/dok2001/status/1937583879986004165 - The peaks and valleys of utilization around the world follows the sun and the moon, but we pay for the power regardless. @eastdakota has sent me that graphs too many times to count over past 13 years. https://t.co/bEDyQVCSlF - **MikeNomitch** (2025-06-24): https://x.com/MikeNomitch/status/1937583820464881740 - @samgoodwin89 Important caveat: internal only... for now... - **duncan** (2025-06-24): https://x.com/duncan/status/1937583726654804389 - This is pretty cool and a great addition to the CF platform. https://t.co/SmFpffCddn - **sargampoudel** (2025-06-24): https://x.com/sargampoudel/status/1937580380238073871 - @irvinebroque @kuldeepsaini_23 Hey just wondering, how useful are containers for building some sandbox type of environment where you let users execute arbitrary code something like a nextjs project - **Just_Moh_it** (2025-06-24): https://x.com/Just_Moh_it/status/1937579637149945949 - this is massive, congrats on the launch, @Cloudflare!! will try building @OpenEditApp on it https://t.co/LCldMcay8t - **eastdakota** (2025-06-24): https://x.com/eastdakota/status/1937578845126668518 - @aniirban101 @Cloudflare @ritakozlov @kristianfreeman It’s meant to fill in gaps that you can’t yet build with Workers. We can offer massively better pricing/scaling/performance with Workers, so want to incentivize workloads run there. Containers are more up to users to optimize, so we can’t be as aggressive on pricing. - **samgoodwin89** (2025-06-24): https://x.com/samgoodwin89/status/1937577841610359194 - Scheduling Policy "moon" to schedule your container to follow the off-peak hours. Now this is where Cloudflare shines. https://t.co/jOt69MuEGT https://t.co/iB6gvDGyBR - **dok2001** (2025-06-24): https://x.com/dok2001/status/1937576767885033518 - Big day for @Cloudflare as we launch our newest compute primitive, Containers! A bit of history: In 2020, we acquired S2 Remote Browser tech and faced the challenge of migrating it from AWS to our edge. To run a Chromium-based browser securely, we split our team: half focused on the Remote Browser app, half built a robust, independent container platform to support it. We knew some workloads needed to be close to users but didn’t fit our Workers isolate model. This platform became a game-changer, empowering dozens of internal teams to build features like Workers CI/CD, Browser Rendering, Key Transparency, Workers AI, and more. But we kept asking: Is this the right primitive for our users? Workers remains the go-to for globally distributed, effortlessly scalable compute at a great price. Initially, many use cases we heard were for single-node webservers that didn’t need region earth. Then we got excited as users started asking for latency-sensitive, real-time applications and the ability to run agents close to the users they serve. Cloudflare Containers are here to deliver for those high-performance, user-proximal workloads. Excited to see what you build with it! - **beyoo94** (2025-06-24): https://x.com/beyoo94/status/1937576063388848426 - What's really cool is that just by deploying to Cloudflare, the image is pushed automatically to their own registry. Also a containers state is handled by using durable object, which is great. - **MrAhmadAwais** (2025-06-24): https://x.com/MrAhmadAwais/status/1937572923822473384 - @kristianfreeman @Cloudflare @CloudflareDev Been using Cloudflare containers for a couple of months now. Pretty happy. Hopefully we can have better limits soon so we can take them in prod. - **threepointone** (2025-06-24): https://x.com/threepointone/status/1937571805213970580 - vibe coded this in a few minutes https://t.co/EnbyAnmFru https://t.co/9hG3ydOiOq - **thomasgauvin** (2025-06-24): https://x.com/thomasgauvin/status/1937568268182585379 - how Workers <> Containers works from @MikeNomitch_CF's demo: 1) pass a fetch from Workers -> Containers 2) fetch from Containers -> Workers/Containers using container-specific domain (🔜) 3) connect Containers -> R2/Hyperdrive using handlers (🔜) https://t.co/FnDW4rR8wQ https://t.co/5A6wNcwYlZ - **TobiM** (2025-06-24): https://x.com/TobiM/status/1937560324598575408 - I'm launching https://t.co/eGCZsqdNuZ as @Cloudflare Containers launched in Public Beta today! If you've build something cool, or wrote a blog post, feel free to open a PR! - **elithrar** (2025-06-24): https://x.com/elithrar/status/1937558668209455591 - @jonas cc @MikeNomitch_CF @irvinebroque - **NatalieMarleny** (2025-06-24): https://x.com/NatalieMarleny/status/1937558308098851087 - Trying it out now! 🫡 https://t.co/VVDDdejOyg - **_ashleypeacock** (2025-06-24): https://x.com/_ashleypeacock/status/1937553724253307064 - Want to get started with Cloudflare's container platform that just dropped? I've got you covered via my FlaredUp newsletter 👇🏻 https://t.co/cR5TBK4gHz - **mwylde** (2025-06-24): https://x.com/mwylde/status/1937552470139605416 - It's live! Serverless pricing, scale-to-zero, and just a wrangler deploy away. https://t.co/WQOLkqSuOf - **eastdakota** (2025-06-24): https://x.com/eastdakota/status/1937551382812217379 - Boom. https://t.co/G0MxiXO4DJ - **zebassembly** (2025-06-24): https://x.com/zebassembly/status/1937551372318036053 - huge huge huge congrats to all my amazing coworkers who were involved in this massive effort. Undoubtedly the biggest Developer Platform release since Durable Objects IMO https://t.co/wEmOTVt0QM - **elithrar** (2025-06-24): https://x.com/elithrar/status/1937550737489428830 - no containers? well, let's fix that: $ npm create cloudflare@latest -- --template=cloudflare/templates/containers-template https://t.co/SRicRF4Kt2 - **TysonTrautmann** (2025-06-24): https://x.com/TysonTrautmann/status/1937550663258636376 - Huge launch! Excited to get your feedback. https://t.co/HecmntaVCc - **kristianfreeman** (2025-06-24): https://x.com/kristianfreeman/status/1937549818861076598 - We're launching @Cloudflare Containers in public beta today. You can take a Docker container and deploy it to our network, with on-demand instancing and load balancing. https://t.co/bY6UaLd0Sd - **juan_ferreras_** (2025-06-24): https://x.com/juan_ferreras_/status/1937549454170890474 - Cloudflare's Containers are now in public beta. https://t.co/M4oGV71axc I was recently struggling to get a WASM version of brotli dictionary_generator (skill issue), but it's a very powerful unlock to just deploy a dockerfile if you need any specific runtime/CPU-intensive work - **andrewe** (2025-06-24): https://x.com/andrewe/status/1937549187266306161 - @jeff_weinstein Cloudflare built such an extensive MCP offering they have 13; granularity helps minimize token count. I use *observability* daily for Cloudflare Workers. https://t.co/46Vxtemrw6 - **rafalwilinski** (2025-06-24): https://x.com/rafalwilinski/status/1937547892371112346 - True Serverless containers-as-a-service with scaling to zero is finally here. https://t.co/PBa3Hpk5di https://t.co/X7DRvLCeXS - **whoiskatrin** (2025-06-24): https://x.com/whoiskatrin/status/1937546573983924254 - containers are out! 💥 oh, and we hid a little surprise in the examples: a coding agent 👀 https://t.co/9BdSnC7ojf - **ritakozlov** (2025-06-24): https://x.com/ritakozlov/status/1937545383112052969 - cloudflare containers are here 👋 https://t.co/ljDGFlJJSv https://t.co/GuqLE2IBeJ - **Cloudflare** (2025-06-24): https://x.com/Cloudflare/status/1937544770752045168 - Cloudflare Containers are now available in public beta. Deploy simple, global, and programmable containers alongside your Workers: https://t.co/L49W1GCVth - **craigsdennis** (2025-06-24): https://x.com/craigsdennis/status/1937541883070403071 - 📦 Containers are here!!! I whipped up was a @Python_dv @FastAPI server that provides shell access to the underlying container's OS. A working sandbox for **ANY** executable you were craving in your existing full stack @CloudflareDev apps! More in 🧵 https://t.co/pQJsWQW7o0 - **RedwoodJS** (2025-06-24): https://x.com/RedwoodJS/status/1937526125426803001 - What makes RedwoodSDK different? It’s not just React on Cloudflare. It’s: 🧠 A config-based router modeled after request/response 🗄️ D1 DB, queues, crons, and auth — all out of the box ⚙️ Transparent DX — no hidden magic Start fast → understand deeply. 👉 https://t.co/6HX5XrrIM3 - **wesbos** (2025-06-24): https://x.com/wesbos/status/1937522820558250019 - MCP Jam is a nice fork of the MCP Inspector - ability to connect to multiple servers at once, save requests and built in chat. Nice little playground to explore existing servers - try the Sentry or Cloudflare ones https://t.co/FbCD8yr7cV - **RedwoodJS** (2025-06-24): https://x.com/RedwoodJS/status/1937488881332498739 - 🚀 Redwood SDK v0.1 is here! The full-stack React framework you've been waiting for: ✅ React Server Components out of the box ✅ Built-in D1 Database ✅ Passkey Auth ✅ Queues and Crons ✅ R2 Storage ✅ Deploy to Cloudflare in one command No magic. No GraphQL. Just a DX-focused full-stack framework. 👉 https://t.co/6HX5XrraWv - **vite_js** (2025-06-24): https://x.com/vite_js/status/1937484461261627681 - ⚡️ Vite 7.0 is out! - Browser Target Changed to Baseline Widely Available - Node 18 support dropped. Vite is now distributed as ESM only - And feats, fixes, and cleanups, while we prepare for rolldown! https://t.co/qGWcL2YQNl - **dok2001** (2025-06-24): https://x.com/dok2001/status/1937403086638891158 - Pricing and billing: the toughest challenges in computer science. https://t.co/b7ZjKQouGr - **Hali** (2025-06-24): https://x.com/Hali/status/1937344254507978938 - Man. Don’t know how to explain it other than shock. Words cannot express the pain of this letdown. The frustration is unfathomable. I’ve worked my whole life to get to this moment and this is how it ends? Makes no sense. Now that I’ve gotten surgery, I wish I could count the number of times people will tell me I’m going to “come back stronger”. What a cliche lol, this shit sucks. My foot feels like dead weight fam. But what’s hurting most I think is my mind. Feel like I’m rambling, but I know this is something I’ll look back on when I’m through this, as something I’m proud I fought through. It feels good to let this shit out without y’all seeing the kid ugly cry. At 25, I’ve already learned that God never gives us more than we can handle. I know I’ll come out on the other side of this a better man and a better player. And honestly, right now, torn Achilles and all, I don’t regret it. I’d do it again, and again after that, to fight for this city and my brothers. For the chance to do something special. Indy, I’m sorry. If any fan base doesn’t deserve this, it’s y’all. But together we are going to fight like hell to get back to this very spot, and get over this hurdle. I don’t doubt for a second that y’all have my back, and I hope you guys know that I have yours. I think Kobe said it best when in this same situation. “There are far greater issues/challenges in the world then a torn achilles. Stop feeling sorry for yourself, find the silver lining and get to work with the same belief, same drive and same conviction as ever.” And that’s exactly right. I will do everything in my power to get back right. My journey to get to where I am today wasn’t by happenstance, I’ve pushed myself every day to be great. And I will continue to do just that. The most important part of this all, is that I’m grateful. I’m grateful for every single experience that’s led me here. I’m grateful for all the love from the hoop world. I don’t “have to” go through this, I get to go through this. I’m grateful for the road that lies ahead. Watch how I come back from this. So, give me some time, I’ll dust myself off and get right back to being the best version of Tyrese Haliburton. -0 Proverbs 3:5-6 “Trust in the Lord with all your heart, and lean not on your own understanding; in all your ways acknowledge Him, and He will make your paths straight." - **theo** (2025-06-24): https://x.com/theo/status/1937329428050375165 - Snapped my board but at least I got a trick https://t.co/iwbDI0C4Eq https://t.co/MHCBywF4Yj - **DKThomp** (2025-06-23): https://x.com/DKThomp/status/1937264368846086207 - Ten years ago, I never would have thought player physical health would be a bigger topic in basketball than football But with almost every recent NBA playoff now marked (and in some cases totally transformed) by lower-body ligament and tendon injuries, it’s kind of happening - **samgoodwin89** (2025-06-23): https://x.com/samgoodwin89/status/1937251836442083444 - Alchemy's infrastructure hot reloading is so fast you can just comment and un-comment the R2Bucket and watch it synchronizes in seconds. https://t.co/YnUSZGbfCZ https://t.co/bkGfI4J1YM - **samgoodwin89** (2025-06-23): https://x.com/samgoodwin89/status/1937244065504055304 - Support for Miniflare with local and hybrid remote is finally available in Alchemy. Hot reloading for all of your infrastructure, and extremely fast ⚡️ https://t.co/4L1kZnvWpN - **wesbos** (2025-06-23): https://x.com/wesbos/status/1937209342802919798 - Currency rates + conversion MCP server is up - both the code and the authless server. Running on Cloudflare workers with their Agents SDK https://t.co/0IEJoMzdoN https://t.co/poO3Ozowp8 https://t.co/THSiYCBGtC - **darraghcurran** (2025-06-23): https://x.com/darraghcurran/status/1937206551447765085 - Picking the best just got easier. We understand that it can be hard to pierce through the noise and with products like @Fin_ai results are what matter most, not marketing or a sales pitch. So we are covering the risk with our Fin Million Dollar Guarantee. https://t.co/PaXXuTYbln - **mwylde** (2025-06-23): https://x.com/mwylde/status/1937195038167224798 - Cloudflare Workers are incredibly powerful, easy to operate, and cheap. But not everything can be written in JS or compiled to WASM. Until now that meant running some applications in another cloud and bridging to CF, but no more! https://t.co/ii57Lf6U8W - **TobiM** (2025-06-23): https://x.com/TobiM/status/1937184839007953303 - .@Cloudflare is cooking something great. I was part of the Containers beta program and I can only recommend to give it a try... https://t.co/0I19jO1kpD - **destraynor** (2025-06-23): https://x.com/destraynor/status/1937182031907340677 - We're putting our money ($1M) where our mouths (or more specifically our resolution rates) are with @Fin_ai - read @eoghan's post for the details. - **eoghan** (2025-06-23): https://x.com/eoghan/status/1937179226383274460 - Today we’re launching the Fin Million Dollar Guarantee. We’ve been making big claims about our industry-best performance, but we know such claims are cheap. So it’s time to put our money where our mouths are. If we don’t deliver on these claims, we’ll pay you $1M. First, some background: In the very near future, all of your customer operations will run on AI. All customer touch points, at every stage in the customer lifecycle, from marketing, sales, service, and success, will be owned by a perfect, charming, expert customer agent, engaging with every customer and available instantly every hour of the day. You’ll choose one vendor for all of this, not many. And you’ll choose the vendor who can do the most high quality work. For the vendors, the opportunity is massive. We’re talking about disrupting literally trillions of dollars of human work, and potentially chasing TAMs 25-50X the size in SaaS. So, unsurprisingly, this space is extremely hot, well funded, and fiercely competitive. For this reason, I’m about to be very honest with my competitive analysis and say the things out loud which every confident CEO says behind closed doors. But please don’t imagine I’m attacking anyone. I know a lot of the CEOs in this space, and admire all of the companies pushing this category forward. The new guys in particular are doing a very impressive job, and it’s generally assumed that they’ll eat the lunch of older incumbents like us. At the risk of over simplification, and to cut to the chase, I believe there are only three players in this space who matter: our product @Fin_ai, and newcomers @SierraPlatform, and @DecagonAI. Sierra is taking a consulting oriented approach with non-tech enterprise customers. This will be extremely lucrative in the near term, but I’m bearish on this approach in the longer term. In the early days of the web, large consulting companies made the exact same enterprise play, charging millions of dollars to deploy and run web sites. But web sites soon became an integral part of business operations and all enterprises eventually wanted to manage them in-house. The same will happen with agents, and in-house teams will need a management application layer, which we’ve made extensive investments in. With Fin we’ve started slightly lower in the enterprise layer, with tech or tech-forward companies who are ready for more sophisticated deployments of AI, which they want to manage themselves. We’re competing directly with Decagon for this market. And while the customers we’re winning against them cite our uniquely deep management application layer, it’s our superior performance that delivers the most striking and immediately valuable difference. It’s become very clear that our AI Group has built highly differentiated tech, which we’ll reveal more about this summer. The result is that we’ve beaten Decagon on performance in 100% of bake-offs that we’ve engaged in and have been chosen as vendor of choice 100% of those times too. Take a representative case of [very large SaaS unicorn], who was an early reference customer of Decagon’s. In a customer-run bake-off, Fin achieved a resolution rate of 63% (which has continued to increase to 72%, as of last week), while Decagon achieved a resolution rate of 49%. That’s 23 percentage points of additional customer requests which now receive instant, perfect resolutions to their problems. And 23 percentage points of additional customer requests that they pay just $0.99 per resolution for, instead of the $15-20 per resolution they’d have to pay humans for. With a performance gap this wide, your choice as a customer is made for you. This dynamic against them, and similar players in the market, has made Fin the largest by customer count (>5k) and largest by ARR (8 figures, which will cross $100M <3 quarters from today) in the category. ARR from our existing customers is growing at 4X, but incredibly, from new business customers it’s growing at 5X. We’re also rated #1 in the category on G2 and shipping every major new innovation in this category before everyone else. So, to the guarantee. Starting today, we promise two things: 1. If you sign up for Fin and are not 100% satisfied in your first 90 days, we will give you up to $1M of your money back, no questions asked. 2. If you sign up for our Fin Guarantee Success Program, and we do not achieve at least a resolution rate of 65% (which our studies have shown is the resolution rate of humans), we will pay you $1M! To learn more about the guarantee and to apply for the program, please visit: fin DOT ai SLASH guarantee :) - **MikeNomitch** (2025-06-23): https://x.com/MikeNomitch/status/1937176648308105326 - Incredibly excited to show off Containers and Workers tomorrow. https://t.co/BUtpgghvpR - **CloudflareDev** (2025-06-23): https://x.com/CloudflareDev/status/1937174722413752530 - 📦 🔜 Join @MikeNomitch_CF and friends at @coderhq @ValDotTown live online tomorrow (June 24) for updates, demos, and AMA. 🔗https://t.co/OyHN2QWwCL https://t.co/hDFPgTXi5k - **ghostwriternr** (2025-06-23): https://x.com/ghostwriternr/status/1937079493144760774 - I’ve joined Cloudflare to build AI agents, and make it the best place for you to build yours! Agents are going to be everywhere, and the team here is full of incredibly talented people building internet-scale infrastructure to make that happen. Stoked to be a part of the band 🕺 https://t.co/gBoY7WqgHH - **DrNiravPandya** (2025-06-23): https://x.com/DrNiravPandya/status/1936997200489529750 - Space out the games in the regular season (and/or decrease the number of regular season games), no back to back games, and increasing the duration of time between playoff games. Also (perhaps most importantly) decreasing the wear and tear that these player have even before entering the league (limit year round AAU at young ages; i.e. 8-9 years old) - **ohnohedidnt24** (2025-06-23): https://x.com/ohnohedidnt24/status/1936995152981287220 - SGA on Hali's injury: "My heart dropped for him. I couldn't imagine playing the biggest game of my life and something like that happening. I just felt so bad for him. Prayers go out to him for sure. Hell of a player" https://t.co/WGqxBJh59l - **DrNiravPandya** (2025-06-23): https://x.com/DrNiravPandya/status/1936993270917431699 - Soft tissue injuries defined these NBA playoffs. Three Achilles tears in the playoffs is not normal. Warriors losing due to Curry's hamstring injury with Denver being impacted by Aaron Gordon's hamstring as well. The NBA has to look at making some changes. - **manuginobili** (2025-06-23): https://x.com/manuginobili/status/1936983292370375154 - Congrats OKC on an amazing season! Well deserved NBA Champ!! 👏👏 Massive respect to the Pacers for being the ultimate example of resilience and courage. Lastly, I wish all the best to Haliburton on his recovery. Very sad to see him going down like that. - **BleacherReport** (2025-06-23): https://x.com/BleacherReport/status/1936970300614472026 - https://t.co/Rrd2ZOcarp - **JeremySochan** (2025-06-23): https://x.com/JeremySochan/status/1936964537032556727 - Can TJ get a foul??? - **VicOladipo** (2025-06-23): https://x.com/VicOladipo/status/1936959776212009011 - In the Finals? So we just going keep playing right? Like its okay right? Like it ain’t happen huh? Like this is just part of it right?But what would this play off run be without Hali? What about Dame what about JT and so many others. I bet we will just blame it on the moment though lol and Hey he got millions right so he will be alright huh? So what truly matters? The Athlete 🤔?. It’s all good though. I promise you im going to be a part of this change. It doesn’t have to be a norm or just part of the game anymore in any sport. We can and will protect the players better. Yall enjoy the rest of the game though. Rest up Hali bro this will make you stronger. 🙏🏿🙏🏿🙌🏿🙌🏿 #HimOlaYa 🤹🏿‍♂️ #ImNotMadImNotUpsetImFocusedOnTheResolution - **CBSSports** (2025-06-23): https://x.com/CBSSports/status/1936958502528077881 - ❤️❤️❤️ https://t.co/RDC2ClplWT - **tomhaberstroh** (2025-06-23): https://x.com/tomhaberstroh/status/1936950583602360644 - Tyrese Haliburton, really a season for the ages https://t.co/YUje7hIUvt - **KarlTowns** (2025-06-23): https://x.com/KarlTowns/status/1936948063018291517 - You never want to see that, prayers up 🙏🏽 - **Carsobi** (2025-06-23): https://x.com/Carsobi/status/1936947233934688431 - This should be a moment of reckoning about the absolute garbage coverage ESPN has delivered throughout an incredible NBA postseason. From shitting on players for playing through injuries to not playing through em, playing Solitaire and all the usual garbage. It's a disgrace. - **lockedupjb** (2025-06-23): https://x.com/lockedupjb/status/1936946467886784856 - Haliburton tearing his achilles in Game 7 of the NBA finals has to be one of the worst things that’s ever happened in sports - **jga41agher** (2025-06-23): https://x.com/jga41agher/status/1936946185564225644 - Tyrese Haliburton is legit one of the nicest, most sincere NBA players I’ve ever met. Just gutted for the guy. - **spidadmitchell** (2025-06-23): https://x.com/spidadmitchell/status/1936946128609808797 - These calf strains ain’t no joke man!! Praying for Ty 🙏🏾🙏🏾 - **jalenbrunson1** (2025-06-23): https://x.com/jalenbrunson1/status/1936945869234032965 - Prayers up man 🙏🏽 - **ramonashelburne** (2025-06-23): https://x.com/ramonashelburne/status/1936944955726938139 - This is awful. Haliburton was playing through the calf injury and playing so well to start this Game 7. I’m sick just watching this. https://t.co/aX3Lyx19jO - **TheTraeYoung** (2025-06-23): https://x.com/TheTraeYoung/status/1936944800005255228 - 🙏🏽🙏🏽🙏🏽 Never wanna see that😔 - **joshhart** (2025-06-23): https://x.com/joshhart/status/1936944344574894323 - Ty 🙏🏽🙏🏽🙏🏽 - **KingJames** (2025-06-23): https://x.com/KingJames/status/1936944290736807993 - FUCK!!!!!!!! 🤦🏾‍♂️🙏🏾 - **KevinOConnor** (2025-06-23): https://x.com/KevinOConnor/status/1936944272583848441 - I'm so sick for Haliburton. This is so horrible. He knew what it was right away. He has the heart of a champion going out there to fight for his teammates. And he was so amazing to start Game 7. Damn it man. - **BaronDavis** (2025-06-23): https://x.com/BaronDavis/status/1936944251558125602 - Maaan my heart goes out to Haliburton. Fck Game 7. Prayers up. - **tomhaberstroh** (2025-06-23): https://x.com/tomhaberstroh/status/1936944155286028547 - Unbelievable. If that's an Achilles tear for Tyrese Haliburton, that's the third one on the Pacers this SEASON. James Wiseman and Isaiah Jackson the others. - **MarcJSpears** (2025-06-23): https://x.com/MarcJSpears/status/1936944144703963370 - Win or lose, Tyrese Haliburton will forever be a Pacers legend for literally putting his body on the line. - **davidaldridgedc** (2025-06-23): https://x.com/davidaldridgedc/status/1936944040362045620 - That is...devastating to watch. - **mrvincecarter15** (2025-06-23): https://x.com/mrvincecarter15/status/1936944033659855289 - Damn man smh 🙏🏽🙏🏽🙏🏽 - **PatrickMahomes** (2025-06-23): https://x.com/PatrickMahomes/status/1936943938327179588 - Prayers up man… 🙏🏽🙏🏽🙏🏽🙏🏽 - **joshgiddey** (2025-06-23): https://x.com/joshgiddey/status/1936943788490138010 - Hali😔😔😔😔 - **PatrickMahomes** (2025-06-23): https://x.com/PatrickMahomes/status/1936943728381354246 - No way man…. - **TheHoopCentral** (2025-06-23): https://x.com/TheHoopCentral/status/1936943707590418583 - Prayers up for Tyrese Haliburton. 🙏 https://t.co/K6sLRF9cjA - **BillSimmons** (2025-06-23): https://x.com/BillSimmons/status/1936943625503719581 - Come on - **ZachLowe_NBA** (2025-06-23): https://x.com/ZachLowe_NBA/status/1936943485979877490 - oh my god - **KevinOConnor** (2025-06-23): https://x.com/KevinOConnor/status/1936943363535904963 - Oh no Haliburton - **MKBHD** (2025-06-23): https://x.com/MKBHD/status/1936940132721889701 - Rooting for the Pacers just for the storyline - **DhravyaShah** (2025-06-23): https://x.com/DhravyaShah/status/1936937924832211104 - WOW you can get a view of your cloudflare workers bundle using `wrangler deploy --metafile=meta.json` then drag + drop meta.json on https://t.co/i0fdzhgzPS enjoy the view :) Thanks @hichaelmart for the tip https://t.co/61lWIuzopO - **DanFeldmanNBA** (2025-06-22): https://x.com/DanFeldmanNBA/status/1936828115877044385 - Suns' depth chart: PG: SG: Devin Booker, Dillon Brooks, Bradley Beal, Jalen Green, Grayson Allen SF: PF: C: - **TyreseHaliburn** (2025-06-22): https://x.com/TyreseHaliburn/status/1936644525859979480 - Today is a great day to win an NBA Championship - **m_franceschetti** (2025-06-21): https://x.com/m_franceschetti/status/1936517184446222769 - @levelsio We will create the @eightsleep hotel and fix the problem - **destraynor** (2025-06-21): https://x.com/destraynor/status/1936461144488804615 - Someone needs to build the GitHub of vibe coding. e.g. Let (normal) people collaborate on a site or app with easy forking, pull requests, rollbacks, and feedback on prompts along the way etc. - **levelsio** (2025-06-21): https://x.com/levelsio/status/1936459630265106931 - I think fitness hotels will be a massive thing in the next few years Every luxury hotel I go asks me "would you like champagne or wine?" during check in Then you enter the room and there's a bottle of champagne waiting Then you go to the restaurant and they give you the 100-page wine list Then you go to the gym and it's 4 treadmills and and no barbell or plates in sight I understand this is because boomers are generally alcoholics, don't work out except walking and they control most of the money; see how Bill Gates looks these days But when they die that will be the largest generational transfer of wealth in history The generations after them are more health focused, don't drink alcohol, want to eat clean and work out properly One hotel I visited last year that fit the bill completely was Siro Fitness Hotel in Dubai (not affiliated, I just like it) They had the best gym I've seen (along with Stay in Phuket which similarly is a fitness hotel, they had a protein shake bar, and the food was all clean meat and vegetables and health-focused Health and fitness is more present in higher income categories so it's perfect for the luxury hotel market I don't see many hotels market to this though, I think there's a lagging demand as they still think the boomers control most of the money, which they do, for now... - **burcs** (2025-06-21): https://x.com/burcs/status/1936290467446243708 - midjourney animations are buttery smooth nothing else I’ve tried even comes close https://t.co/Zx6nvq5QIy - **saranormous** (2025-06-21): https://x.com/saranormous/status/1936262075321663666 - We should celebrate hacking, not cheating. Cheating quietly pockets extra Monopoly money (insider trading, plagiarism, inflated metrics). Hacking openly flips the game board, exposing rigged rules and inventing a better game for everyone: Wikipedia replaced Britannica, Airbnb monetized spare rooms, Linux openly challenged Windows. Both break the rules, but hacking doesn’t compromise integrity. it reveals the old game wasn’t worth playing. - **samgoodwin89** (2025-06-21): https://x.com/samgoodwin89/status/1936259806505189754 - IaC is made easier when APIs support reading back all written data. Eg with CloudFlare, it would be helpful to read the previous migrations so that a migration plan can be automatically derived. Adopt the principle of: > “read what you write” - **FastbreakHoops5** (2025-06-20): https://x.com/FastbreakHoops5/status/1936201189273501703 - Nahh, these third graders acting different now 🤣 https://t.co/exK1tZz8Ob - **korinne_dev** (2025-06-20): https://x.com/korinne_dev/status/1936192070261305557 - @mattzcarey @CloudflareDev We actually just released support for this! https://t.co/J9Ppj3orDK - **destraynor** (2025-06-20): https://x.com/destraynor/status/1936146578638586108 - Started a new podcast with @Padday — I promise I’ll never be this hairy on camera again (I didn’t expect it to air 😅) We covered: 1 Don’t build a MiniDisc 2. AI= both new market + low-end disruption 3 Roles are collapsing 4 UIs are rebundling 5 Cynics will be left behind Enjoy https://t.co/Bk7UgQfPoo - **KentonVarda** (2025-06-20): https://x.com/KentonVarda/status/1936146315429220364 - I did a podcast thing! Talking about AI-assisted coding and Cloudflare Workers. Hot take: With AI there will be more developers, not fewer, and a lot of them will be building on Workers. Please watch this teaser while I go find my flame-retardant suit. (Full vid in reply.) https://t.co/GVuuIm63tl - **ritakozlov** (2025-06-20): https://x.com/ritakozlov/status/1936136541018968468 - it’s ship summer at @cloudflare this is such an awesome improvement! i never want to manually curl another multipart upload again 💆‍♀️ https://t.co/aHi6aG00sT - **Cloudflare** (2025-06-20): https://x.com/Cloudflare/status/1936134832494461255 - When AI writes code — challenges, opportunities, and the need for guardrails. @KentonVarda, Principal Engineer at Cloudflare, talks trust, review, and how Workers (@CloudflareDev) can scale AI-assisted dev while making it safe. 🎙️ Watch #ThisWeekInNET: 🔗 https://t.co/ArpcS6CSpt https://t.co/q1x7agG0Og - **korinne_dev** (2025-06-20): https://x.com/korinne_dev/status/1936132690962792526 - We've abstracted away the complexity of `multipart/form-data` uploads in our Cloudflare SDKs! Plus a bunch of fixes to our Terraform provider. https://t.co/IIbx4lwamA - **elithrar** (2025-06-20): https://x.com/elithrar/status/1936109516569059746 - TWO of my favorite people (@dinasaur_404 *and* @cjbell_) at an MCP server meetup?! https://t.co/bO7FFAn0AS - **elithrar** (2025-06-20): https://x.com/elithrar/status/1936093563500065226 - we call this a "mike drop" now https://t.co/AYsilSNK4Z - **yusukebe** (2025-06-20): https://x.com/yusukebe/status/1936089907950502184 - 今honoとastroならhonoの方がダウンロード多いのか。これはすごい https://t.co/AWiGiJzfNo - **elithrar** (2025-06-20): https://x.com/elithrar/status/1936086476283924895 - in our weekly @cloudflare Workers team demo session where @xortive is showing off how he builds (complex!) Grafana dashboards on top of our internal analytics/metrics schemas using Claude Code. saving literal hours of writing queries panel by panel. - **aninibread** (2025-06-20): https://x.com/aninibread/status/1936077351974031595 - Also, small but mighty update: AutoRAG now supports PDFs up to 4 MB (previously 1 MB)! https://t.co/kpCmQnza3R - **aninibread** (2025-06-20): https://x.com/aninibread/status/1936069554976796871 - Three new drops in AutoRAG this week: 🔍 Filter by filename to control what gets retrieved 🧠 Add a context field to enrich each chunk 📎 See custom metadata in search responses Docs → https://t.co/R72QfDJDId What do you want to see next? 👀👀 https://t.co/mLiPJAutA9 - **janwilmake** (2025-06-20): https://x.com/janwilmake/status/1936059017131917751 - This is such a brilliant blog, I think I've read it 4 times now. The system behind DO's with SQLite is just brilliant. https://t.co/Qgj04rUqce - **DKThomp** (2025-06-20): https://x.com/DKThomp/status/1936051079239668202 - Some personal news. Today, I’m leaving The Atlantic after almost 17 years and moving my writing to Substack. It would be convenient, for the purposes of crafting an exciting departure announcement, to have a dramatic exit story: a fight, a grievance, a shouting match with an editor that ended with me hurling a bunch of leather-backed Thoreau volumes across the open-plan office. That is not the case here. I love The Atlantic, and I'll remain a contributing writer there. But after almost two decades at one publication, I wanted to write for myself. The things I've published that I'm most proud of—whether it was the original abundance agenda essay, or my piece on workism—emerged from a very personal expression of frustration, or confusion, or curiosity. I want to know what my thinking and writing is like if I lean into a more independent and personal writing life. That's brought me to Substack, which is already home to an astonishing share of my overall reading. I'm excited to join their community and excited to build my own. The name of the newsletter should be easy to remember: Derek Thompson. The newsletter will have three main pillars 1. Abundance 2. The frontier of science and technology—GLP1s, AI, biotech, energy breakthroughs—covered in a way that’s both curious and skeptical 3. The anti-social century & the social crises of anxiety and aloneness Thanks to The Atlantic for 16.8 incredible years and thanks to everybody who follows me across the river. - dt - **wesbos** (2025-06-20): https://x.com/wesbos/status/1936050634110697497 - I let cursor rip overnight on some large tasks and I came back to it having the same struggles we all have https://t.co/6nIybACuDx - **liadyosef** (2025-06-20): https://x.com/liadyosef/status/1936038502162219040 - An interesting update recently - an MCP workgroup on UI integrations was formed! There are open, community-led discussions about how the standard should treat UI, and this workgroup is aimed on focusing around these questions. @idosal1 and I are moderating it. - **StartupArchive_** (2025-06-20): https://x.com/StartupArchive_/status/1936028585384395043 - Alexandr Wang: “Hire people who give a shit” The ScaleAI founder is asked for the best advice he has for founders, to which he responds: “The biggest thing is you have to really, really care.” Alex explains this is also the number one trait he looks for in people he hires: “When you interview and interact with people, you can tell who phones it in and whose work is so incredibly important to them that it eats at them if they don’t do great work. One of the greatest indicators of how much I enjoyed working with people and how successful they were at Scale was just the degree to which their soul is invested in the work they do.” It’s important that the CEO lives this principle too. “I still review every hire at the company,” Alex says. “I care immensely. I work with all these people who care immensely. And that lets us feel much more deeply what happens in the business, so we learn more quickly and adapt more quickly, and I think that’s been quite important to the success that we’ve had.” One of Scale AI’s values is that “quality is fractal.” “High standards trickle down within an organization,” Alex explains. “Most of the time when people realize their manager or their manager’s manager doesn’t really care, that removes the desire for them to care. So it’s incredibly important that high standards and care for quality is a deeply embedded tenet of the entire organization.” Video source: @ycombinator (2025) - **NBA** (2025-06-20): https://x.com/NBA/status/1935887401601368204 - PACERS BASKETBALL. IT'S INCREDIBLE. https://t.co/t19RJrgJSy https://t.co/MNT3QrwpDo - **threepointone** (2025-06-20): https://x.com/threepointone/status/1935859748210897271 - "call my agent" --- tl;dr - you can make phone calls to durable objects! and agents! I ported over the openai ⨉ twilio example and ran it inside a cloudflare agent, and it just works! https://t.co/lis6thzi5l - **kathyyliao** (2025-06-19): https://x.com/kathyyliao/status/1935845386809491477 - Generate pdfs, take screenshots, grab markdown + more. Just one api call away. Let me know what actions you’d like to see next~ https://t.co/yVZpHJeGEa https://t.co/XUbYH221PP - **ryolu_** (2025-06-19): https://x.com/ryolu_/status/1935842676806762679 - @Stammy @msfeldstein worked on this beauty ahead of its time - **G4brym** (2025-06-19): https://x.com/G4brym/status/1935836437158527270 - You can now deep research your Cloudflare AutoRAG in workers research You can also research the web + AutoRAG for better results Just pick an existing RAG when creating a new job and that's it! https://t.co/R1Zoas1S27 - **Stammy** (2025-06-19): https://x.com/Stammy/status/1935831067745042756 - people forget how influential Palm webOS was. lots of modern iOS uses similar patterns. palm was basically the OpenAI of their time. such talent density that Steve Jobs had to threaten them to stop poaching. • horizontal app card multitasking w/ flick up to quit • edge swipe gesture nav for back nav and quick app switching • swipe away banner notifications • swipe down system tray (like Control Center) • "Just type..." global search bar • threaded messaging • magnetic charging puck called Touchstone (like MagSafe) - **kristianfreeman** (2025-06-19): https://x.com/kristianfreeman/status/1935813485025546266 - Generating PDFs is hard. Including puppeteer in your app bundle is error-prone outside of Javascript apps. Use Cloudflare's Browser Rendering API. You can add auth, deploy it, and use it to render PDFs across projects. https://t.co/J5Qc0pgLGk - **thdxr** (2025-06-19): https://x.com/thdxr/status/1935801226362302730 - opencode rewrite is done and ready for general use - works with claude pro/max - beautiful themeable tui - shareable links for any session - zero config LSP support - works with 75+ LLM providers (including local) link in reply https://t.co/R8xhcWTOxj - **ritakozlov** (2025-06-19): https://x.com/ritakozlov/status/1935797565787689017 - one of my favorite new cloudflare products is autoRAG! now with two new additions: 🔸 support for multi-tenancy through custom metadata on search 🔸 ability to pass additional context to the LLM https://t.co/ZuwcxHXGtf - **eastdakota** (2025-06-19): https://x.com/eastdakota/status/1935772589055787380 - I did in fact say this: “I go to war every single day with the Chinese government, the Russian government, the Iranians, the North Koreans, probably Americans, the Israelis, all of them who are trying to hack into our customer sites. And you're telling me, I can't stop some nerd with a C-corporation in Palo Alto?” - **dok2001** (2025-06-19): https://x.com/dok2001/status/1935769244580610552 - A great run! Excited to see what you do next! https://t.co/e2UiLnaUup - **zaid** (2025-06-19): https://x.com/zaid/status/1935768842661224661 - After 7 years, I’m going from working at Cloudflare to building on Cloudflare! Tomm is my last day I’ll miss working with so many insanely driven ppl who made each day fun—none more than @rrnn & @KankaniAbhishek. Thanks @dok2001 for the life-changing DM that started it all 🧡🚀 https://t.co/oop4gX0z9H - **whoiskatrin** (2025-06-19): https://x.com/whoiskatrin/status/1935743491126313374 - cloudflare is really doing to AI what they did to internet: taking something complex and making it just work everywhere and for everyone - **dsp_** (2025-06-19): https://x.com/dsp_/status/1935740870680363328 - We released a new MCP specification: 2025-06-18. It contains a lot of improvements over the previous version: We improved how authorization works making it easier for enterprises to build with MCP, added elicitation and resource links and more: https://t.co/pLpM2LAVsp - **Esteban_Puerta9** (2025-06-19): https://x.com/Esteban_Puerta9/status/1935727010480865402 - I’m so sold on @Cloudflare ‘s entire developer platform https://t.co/Szm4ZALvTV - **TobiM** (2025-06-19): https://x.com/TobiM/status/1935650605176942658 - Pretty neat way to integrate remote MCP servers in your React app with a few lines of code: https://t.co/C0abmUdR2B - **RozenMD** (2025-06-19): https://x.com/RozenMD/status/1935624002631856570 - https://t.co/DXPPeWfeSk https://t.co/Cx73LtMldK - **janwilmake** (2025-06-19): https://x.com/janwilmake/status/1935572874560979092 - Huge new feature! You can now develop locally with bindings that live remotely (in staging or even prod) You don't know how often I've just deployed new workers to prod just to test because the data was there and I didn't feel like creating a replication locally. This changes today. I can just bind my remote KV, R2, or D1, while changing my code in localhost. Huge!!!! - **mrexits** (2025-06-19): https://x.com/mrexits/status/1935554373259595991 - Sam Altman on how he decides to invest in a startup in just 10 minutes https://t.co/3YM0zWgj2v - **tobi** (2025-06-19): https://x.com/tobi/status/1935533422589399127 - I really like the term “context engineering” over prompt engineering. It describes the core skill better: the art of providing all the context for the task to be plausibly solvable by the LLM. - **lizziepika** (2025-06-19): https://x.com/lizziepika/status/1935508272389431451 - Here tonight speaking at the June MCP builder series with @apollographql! 🧵 https://t.co/37j1NrffK0 - **lucas__crespo** (2025-06-19): https://x.com/lucas__crespo/status/1935505758294045180 - ChatGPT knows a lot about your work, remembers your preferences, and could easily check in on things you care about. But it just sort of sits there, waiting, like a super polite guest who won't speak until spoken to. This is probably one of the most interesting design challenges right now. We have the most incredibly capable AI that could proactively help us, but it's not designed to be that way. We always have to make the first move. Imagine if ChatGPT could say "hey, remember that research project? Here's new data that changes everything" or "how are you feeling today? you mentioned you had a headache yesterday" This wouldn’t be a simple notification, it would be a moment of shared awareness, the AI noticing something, remembering the context, and choosing when to say something that feels timely and personal. The challenge doesn't seem technical, it's social. There's this whole emotional intelligence layer to it. When does helpful become annoying? How do you build something that can "read the room"? This goes way beyond traditional UX into something more like social design. Tools won't just need better interfaces, they'll need better instincts. - **ycombinator** (2025-06-19): https://x.com/ycombinator/status/1935496106957488566 - Andrej Karpathy's (@karpathy) keynote yesterday at AI Startup School in San Francisco. https://t.co/UM1wfFs98S - **thdxr** (2025-06-18): https://x.com/thdxr/status/1935450224417218991 - why does everyone make a big deal about "hockey stick growth" literally every one of my products does that https://t.co/ZOfMcLjVvh - **glenmaddern** (2025-06-18): https://x.com/glenmaddern/status/1935438933271396551 - 👩‍🍳 https://t.co/8YLwDKEKgb - **thdxr** (2025-06-18): https://x.com/thdxr/status/1935437176122015863 - this means `sst dev` for cloudflare can finally be good https://t.co/l0cW3HLkD8 - **wesbos** (2025-06-18): https://x.com/wesbos/status/1935436237763957222 - Major pain point in Coudflare workers solved - access remote bindings locally. Key if you want to be able to access your prod DB locally, or during CI https://t.co/FAIi38sXrE - **bascodes** (2025-06-18): https://x.com/bascodes/status/1935432721209147485 - @korinne_dev This is huge. This whole bindings thing is probably the biggest selling point for cloudflare. Will try this on the weekend… - **TobiM** (2025-06-18): https://x.com/TobiM/status/1935427396468977727 - Just updated my project for running DuckLake on @Cloudflare Containers to @duckdb v1.3.1: https://t.co/H0dRYYKQL7 - **charl_dot_dev** (2025-06-18): https://x.com/charl_dot_dev/status/1935413908694880359 - Cloudflare Realtime UI Kit is epic, like 10 lines of code and full on conferencing https://t.co/LK7p8q80VV - **korinne_dev** (2025-06-18): https://x.com/korinne_dev/status/1935412758868279462 - Public beta for remote bindings during local dev is out today 🧡 https://t.co/yXvaGyFh1y You can specify which bindings should connect to a deployed resource, and your Worker still executes locally. https://t.co/byDQKgQ4Mb - **dinasaur_404** (2025-06-18): https://x.com/dinasaur_404/status/1935410961118368070 - Thank you @auth0 for including me (and my MCP earrings) in your Dev Day! Hope everyone watching goes and deploys a remote MCP server right now 🙂⚒️ https://t.co/swOsGN2BdE - **auth0** (2025-06-18): https://x.com/auth0/status/1935409509058400419 - WOW! @dinasaur_404 from @Cloudflare just wrapped a 🔥 session on building Model Context Protocol (MCP) servers. 💡 Best practices 🛠️ Real-world patterns 🤖 Agent-ready infra tips ✨ Plus… iconic MCP earrings. If you missed it, catch the replay — this one was gold. https://t.co/nYSYhJSkar - **whoiskatrin** (2025-06-18): https://x.com/whoiskatrin/status/1935393601497792736 - full house of amazing builders in London 🧡 https://t.co/PKUrVFMvOF - **Baconbrix** (2025-06-18): https://x.com/Baconbrix/status/1935377776984420708 - I’ve been a huge fan of Nishan since he added flex gap and CSS gradients to React Native! He opened a PR for ‘filter: blur()’ style support in React Native on Sunday 🫧 @Expo apps are about to get a lot more beautiful 🔥 - **AnthropicAI** (2025-06-18): https://x.com/AnthropicAI/status/1935367951542280239 - Claude Code can now connect to remote MCP servers. Pull context from your tools directly into Claude Code with no local setup required. https://t.co/GUZjQ6Jkbj - **dsp_** (2025-06-18): https://x.com/dsp_/status/1935353293875921369 - Glen is cooking. Glen usually cooks well… https://t.co/TB3UCdebPR - **glenmaddern** (2025-06-18): https://x.com/glenmaddern/status/1935326078459154820 - Announcing use-mcp, a hook for making any React app an MCP client: https://t.co/R20HL60MsN It supports both SSE and Streamable HTTPs remote servers, handles the OAuth flow, should just drop in and give you a list of tools and a way to call them (for now, lots more to come). - **dinasaur_404** (2025-06-18): https://x.com/dinasaur_404/status/1935324944843321688 - Huge thanks to @glenmaddern and @threepointone for making this happen! 👏🧡 https://t.co/W8n2YaVt4q - **CodingGarden** (2025-06-18): https://x.com/CodingGarden/status/1935324587203653904 - In my latest video over on @SyntaxFM I showcase the 9 new features added to the JavaScript language in this years spec. Watch 🍿: https://t.co/xvPugfYhwQ - **Cloudflare** (2025-06-18): https://x.com/Cloudflare/status/1935322874841096319 - We're open-sourcing use-mcp, a React library that connects to any MCP server in just 3 lines of code, as well as our AI Playground, a complete chat interface that can connect to remote MCP servers: https://t.co/NcolMhWvbZ - **dinasaur_404** (2025-06-18): https://x.com/dinasaur_404/status/1935322850723832183 - Two new gifts from @Cloudflare for the MCP community: 🎁 use-mcp: a React hook that makes it easy to connect any web app to remote MCP servers in just 3 lines of code 🛝 AI Playground, open-sourced: a chat interface with built-in LLMs + MCP support that you can deploy yourself! https://t.co/hAh2ft52LP - **brandinpodziem2** (2025-06-18): https://x.com/brandinpodziem2/status/1935195498429956412 - Whose Valkyrie jersey should I wear Thursday 👀 - **reed** (2025-06-18): https://x.com/reed/status/1935184239945007306 - first @waymo @flysfo connection? https://t.co/FamVrGHrUi - **samgoodwin89** (2025-06-18): https://x.com/samgoodwin89/status/1935183889385078952 - Look how neatly Alchemy's IaC integrates with JS frameworks, e.g. OpenAuth. Can instantiate a KVNamespace and pass it directly to the CloudflareStorage. https://t.co/UJMHauF5S2 - **src_rip** (2025-06-18): https://x.com/src_rip/status/1935183672618983697 - @irvinebroque @dok2001 I think I would’ve figured it out a lot faster if the tool tip was actually like in the body of the form, because once I read the tool tip, I kind of figured out that I needed a Wrangler CLI command in there, but I didn’t think to read the tool tip immediately - **src_rip** (2025-06-18): https://x.com/src_rip/status/1935183401943810485 - @irvinebroque @dok2001 There was also some kind of bug where I couldn’t actually submit nothing for the fields marked optional in the build settings. Like it says that the version update command is optional, but I cannot leave it nil, if I do, I get an error that says HML body invalid or something - **src_rip** (2025-06-18): https://x.com/src_rip/status/1935183178748104917 - @irvinebroque @dok2001 I don’t remember exactly but I’m pretty sure when I first linked the GH repo to the worker instance in the ui that field was blank… I could be misremembering but I think it was blank - **fat** (2025-06-18): https://x.com/fat/status/1935141166259064881 - Pretty much all software tool adoption right now is fear based: "YOU BETTER STOP USING VIM AND START USING CURSOR MAX YOU ANALOG PIECE OF SHIT, OR YOU'RE GOING TO LOSE YOUR JOB IN THE NEXT 2 WEEKS". Interesting times… - **yagiznizipli** (2025-06-17): https://x.com/yagiznizipli/status/1935102093699973342 - Soon, node:http and node:https modules are coming to @Cloudflare workers! https://t.co/BJumA4MH8V - **paultoo** (2025-06-17): https://x.com/paultoo/status/1935081914429489280 - Not many people know this, but OpenAI was actually started at YC in 2015, and was initially known as "YC Research". We were concerned that all of the AI technology would be locked up inside of one company (Google), and so created a "moonshot" effort to make the technology available to all, including future YC companies. It's difficult to convey how crazy and unlikely this all seemed 10 years ago, but here we are :) - **saranormous** (2025-06-17): https://x.com/saranormous/status/1935032358522208360 - the power/battery of today’s smartphones is just not gonna work if we really begin to use local models all the time (or even just constant input, eg voice processing) - **reed** (2025-06-17): https://x.com/reed/status/1935014303226380357 - waymo LA now to silverlake, inglewood, and into the hills @waymo https://t.co/0NMP051ZMv - **KevinOConnor** (2025-06-17): https://x.com/KevinOConnor/status/1934810551692988581 - The Jalen Williams Game. Truly incredible. Did literally everything. Becoming the co-star for SGA that OKC needed right before our eyes. - **CloudflareDev** (2025-06-16): https://x.com/CloudflareDev/status/1934757196174172561 - Big fans of what @SST_dev is up to here: https://t.co/O7I2DXkNDH - **thdxr** (2025-06-16): https://x.com/thdxr/status/1934743296347156510 - btw we've been building all our latest stuff on cloudflare and improving sst's support for it as we go here's the code to deploy modelsdotdev https://t.co/1il8SKKrPL - **eastdakota** (2025-06-16): https://x.com/eastdakota/status/1934740692204765614 - Yup. Biggest inhibitor to AI isn’t technical, it’s expectations. We expect computers to be perfect. These new machines are, by design, not perfect but they’re better than humans. For adoption, we need to adjust expectations. That’s a harder problem than if it was just technical. https://t.co/xzK73rmRgm - **jquinonero** (2025-06-16): https://x.com/jquinonero/status/1934716214473048438 - In another unconventional career move, I’m now the Head of Recruiting for all OpenAI. Making this decision was hard and easy.  I care immensely about culture and about people, the further I get in life, the more I understand these are the two most important things. 1/n https://t.co/YJClvj2DxK - **CloudflareDev** (2025-06-16): https://x.com/CloudflareDev/status/1934713365924024508 - no limit to what you can use DOs for 🚀 https://t.co/063TwubYS9 https://t.co/1eCNaQG2tI - **jhomra21** (2025-06-16): https://x.com/jhomra21/status/1934706053708697716 - WIP - Use @convex as main database, have @CloudflareDev Durable Object per user. Write to Convex, read from DO. If a users DO is deleted, its data remains in Convex and can be recovered! https://t.co/3UOhE2sG2M - **polar_sh** (2025-06-16): https://x.com/polar_sh/status/1934681237567721489 - accepting payments with @honojs has never been easier ✨ https://t.co/KI6NVFTdUJ - **mjackson** (2025-06-16): https://x.com/mjackson/status/1934630322802720907 - We used to have a method called componentDidMount. Now it’s useEffect(() => { // … }, []) The kids would never believe it. - **SullyOmarr** (2025-06-16): https://x.com/SullyOmarr/status/1934630067667374152 - i think we got "ai memory" all wrong everyone is focused on making models remember facts about you when its way more useful in a work setting ex: when you explain something to a coworker it's basically: - the task (this is what we do now with prompts, memory has very little importance) - how to do the task -> this is where "raw intelligence" matters (reasoning etc) and ability to follow instructions - where to get the required information (context) - this is the important and what everyone is missing about memory e.g (look at notion for HR, and then check slack and linear for xyz) with any agent right now it can do points 1 and 2 pretty well, but fails at 3. you have to explain the same thing over and over each time. its not figuring how do the job better over time we need to solve for an agent doing a task and remembering exactly how it did, what it learned, etc eg: once it figures out that revenue data lives in stripe → sheets → mixpanel, it should just know that for every interaction onwards - **mjackson** (2025-06-16): https://x.com/mjackson/status/1934629080462418128 - I want a component model that feels native to the web. But I also want to take the best of React. Web components and React have always been at odds. The rendering model. The development and deploy story. Reacts lifecycle used to mirror the DOM. It’s much more obscured behind hooks. But it’s still there. How can we bring it front and center again? How can we make something that feels web native that we can also server render without jumping through hoops? - **glcst** (2025-06-16): https://x.com/glcst/status/1934609480714461404 - We at @tursodatabase have just done something unheard of in the startup world: we hired a person in prison. Not an ex-con: he is actually serving time right now. If you follow us you may remember when we announced our full rewrite of SQLite in Rust, Project Limbo. It got a tremendous amount of attention, but by far, my favorite story was that of @PThorpe92: Preston is currently incarcerated for things he regret doing a decade ago. A model reformed inmate, Preston was given the beautiful opportunity by the Maine Department of Corrections to access the internet. He then found our project, and since it's not like there are many entertainment options competing for his attention in prison, he decided to pour his heart and soul into helping us rewrite SQLite. I immediately fell in love with his story. Reading his old blogs, his Github profile. I prayed we would be able to hire him as soon as he was out, and make him a part of our team. But thankfully, things have happened in such way, that we were able to do it *even before* he was released. He's now given an opportunity to make a dent in the world, from prison. Preston claims he is immensely thankful for this opportunity. It is certainly not an opportunity that comes often, and I understand him. But when I read his story in his personal blog, where he mentioned how he clearly saw his life going downhill, had this sudden epiphany, saw clearly that he could do differently, and then started being showered with blessings, I saw the unmistakable markings of how our Lord Jesus Christ, the Lamb of God, operates in the world. I was an obnoxious Richard Dawkins-style atheist for more than two decades. I am not an experiential person, and I came to a change through the rational evaluation of the historical context of His resurrection. I am not the kind of person that feels the hand of God in my life, but over the years I learned to recognize his patchwork through those sudden, unexpected changes, where everything suddenly "just works". And because of that, I am the one who's thankful. The Lord decided to operate such a drastic change in Preston's life. Clearly He has a great plan for him. I am immensely thankful for the opportunity to play a small part in that story. God, in his infinite power, doesn't need me to do any of that, and yet he gave me the opportunity to act. I feel thankful and blessed. I am also very thankful to @LabsUnlocked , and entity helping people like Preston find a better life once they're out. If you feel inspired by their mission, you should reach out and help. And if you want to hear from Preston in his own words, read the article he just published on the Turso blog: https://t.co/xvp3j0Tnnw - **CloudflareDev** (2025-06-16): https://x.com/CloudflareDev/status/1934600866620957167 - R2 Data Catalog (Iceberg!) 🤝 Parquet in R2 🤝 @marimo_io https://t.co/pjZGDWWLar - **elithrar** (2025-06-16): https://x.com/elithrar/status/1934580655230214513 - IP addresses are not identifiers. IP addresses are not identifiers. IP addresses are not identifiers. IP addresses are not identifiers. IP addresses are not identifiers. IP addresses are not … https://t.co/lkXjWooWav - **marimo_io** (2025-06-16): https://x.com/marimo_io/status/1934561510283706661 - We got three words for you: zero egress fees! We triple checked the docs to make sure we read it correctly, but the new R2 data catalogue offering from @Cloudflare gives you Apache Iceberg with *zero egrees fees*. It even integrates directly with marimo! https://t.co/5HXBoV1lXN - **tobyhardtospell** (2025-06-15): https://x.com/tobyhardtospell/status/1934381442404172176 - @CohenSite Sentence him to building 3,257 housing units in Burlingame https://t.co/7ADPn5SwH9 - **Bob_Wachter** (2025-06-15): https://x.com/Bob_Wachter/status/1934070216327376953 - There are few in this world who were such forces for good as @atulbutte. His dazzling intellect, relentless curiosity, generosity of spirit, boundless charisma, and infinite optimism left a mark on everything he touched and everyone lucky enough to know him. An enormous loss. https://t.co/qN3GYWyEIR - **chiragjp** (2025-06-15): https://x.com/chiragjp/status/1934066719393284529 - Heartbroken to share my mentor & friend Dr. Atul J. Butte @atulbutte has passed. He saw potential in me and turned it into opportunity-- for me, and so many others. We bonded over Star Trek and he was our Captain. Picard’s parting toast in his honor, final episode: “....five-card stud, nothing wild… and the sky’s the limit.” https://t.co/4PkqfomQXb - **garrytan** (2025-06-14): https://x.com/garrytan/status/1933997295667916939 - Software and technology is a force-multiplier for human creativity, not a replacement for it. https://t.co/y6VLJ6g2d4 - **peteratmsr** (2025-06-14): https://x.com/peteratmsr/status/1933979060935483570 - Mourning the passing of @atulbutte. I learned so much from the precious few chances I had to be with him. He's someone who always picked up the phone whenever I needed advice, on anything, anytime. A great man who made everyone around him better. - **banteg** (2025-06-14): https://x.com/banteg/status/1933967436459503662 - guys? https://t.co/MVc93ZJaUJ - **dvir_a** (2025-06-14): https://x.com/dvir_a/status/1933942180625666288 - Between alarms and sheltering, I learned last night that my postdoc mentor, Professor Atul Butte @atulbutte, has passed away, and he is only 55 years old. This is devastating news 💔. Atul was a giant in so many ways. He was a visionary leader in translational bioinformatics @Stanford who reinvented himself as a pioneer of AI in medicine @UCSF. He published hundreds of papers, founded multiple companies, was elected to the National Academy of Medicine, and recently to the American Academy of Arts & Sciences. But Atul was much more than his achievements. He possessed something rare in science - infectious positivity. Every encounter with him, no matter how low I felt going in, I'd leave optimistic and energized to pursue my goals. After every talk he gave, students would line up to tell him how he motivated them to do something big with their lives. Another rare trait was his kindness and generosity with credit. Atul taught me: "credit is indefinitely divisible." In a field where people often take undeserved credit, this was inspiring. I will miss Atul, like so many others will. He is survived by his wife Gini and daughter Kimi, but his legacy will live on through his countless trainees and colleagues worldwide. - **MathurAditya7** (2025-06-14): https://x.com/MathurAditya7/status/1933877635790111222 - Connecting @honojs to your MCP server just got a lot smoother ✨ You can run this on Cloudflare, Bun, Deno, basically all the places where hono can run! I made a video walking through hono/mcp in action - **tweetsbycolin** (2025-06-13): https://x.com/tweetsbycolin/status/1933657833318576204 - Small aside: our team really crushed this launch. We were very much on our backfoot when MCP landed. We supported a flavor of oauth for syncing sessions (e.g. to a Zendesk support panel), but not for broader authorization delegation. There was no consent screen. We also didn't have the infrastructure to offer stateful authentication tokens. Our web auth uses JWTs, which could work, but would be downright unusual for oauth. Super delighted with how quickly this came together, especially given how quickly MCP and agent auth continues to evolve. So much to do! If you're interested in MCP, agent identity, etc - please get in touch through our website /careers! - **tweetsbycolin** (2025-06-13): https://x.com/tweetsbycolin/status/1933653599747518486 - Introducing: const { userId } = await auth({ acceptsToken: "oauth_token" }) Overnight, MCP's spec for agent auth created near-ubiquitous demand for a built-in OAuth Provider Now you can offer an OAuth Provider in a few clicks, AND easily build HTTP handlers for access_tokens https://t.co/pVb8xZZfSv - **leo** (2025-06-13): https://x.com/leo/status/1933560959982805197 - Blade can now be deployed to @cloudflare and @netlify with zero config, in addition to @vercel and containers! 🎉 Let me know if there are any other providers you'd like to see added. I think @deno_land and @awsamplify would make sense? - **destraynor** (2025-06-13): https://x.com/destraynor/status/1933554110575300829 - We're taking a simple idea, and we're taking it seriously. We're aiming to 2x our productivity with smart deliberate adoption of AI across the board in our R&D org. @darraghcurran is driving this effort, and we're sharing his email to the team. (🔗in🧵) https://t.co/2qS4WJtjMf - **darraghcurran** (2025-06-13): https://x.com/darraghcurran/status/1933497547663782139 - Continuing on the momentum @eoghan shared earlier this week - our whole R&D org are taking on a goal to 2x our productivity through mastery of the wonderful new tools and capabilities AI gives us builders. Buckle up! https://t.co/i8rE1wXfZQ - **samlambert** (2025-06-13): https://x.com/samlambert/status/1933330855545721155 - blame vs accountability https://t.co/Z9UwUrQPKV - **dok2001** (2025-06-12): https://x.com/dok2001/status/1933272215396688124 - @altryne @Cloudflare At the end of the day we are responsible for our availability. - **dok2001** (2025-06-12): https://x.com/dok2001/status/1933270392656765397 - We let down our customers at @Cloudflare today. Our Workers KV service failed and the downstream products that rely on that service had outages of their own. We will publish a full postmortem soon. I know that these kinds of incidents have real and serious impact for teams around the world as they do their work and serve their own customers. Beyond the specifics of this incident, the postmortem will go into detail about how we are removing this failure case. We will do everything we can to make this right going forward and I apologize for the problems caused today. - **zebassembly** (2025-06-12): https://x.com/zebassembly/status/1933230858904728041 - so is everyone else having a very calm and totally relaxing afternoon? - **threepointone** (2025-06-12): https://x.com/threepointone/status/1933203542824661167 - I work with a brayden brandon and brendan and I'm having to lay my indian accent on realllly thick so it's clear who I'm talking about - **eastdakota** (2025-06-12): https://x.com/eastdakota/status/1933141676509306940 - Actually, this is a recipe for a really rewarding life if you’re a recent college grad not sure what to do: Find a small town you could love with a local newspaper whose owners are ready to retire. Raise the capital to buy it. Run it with the community’s interests at heart. You’ll not get rich, but you’ll do well. More importantly you’ll be a hero to the community and have influence even early in your career. You’ll meet the love of your life at some event you wouldn’t otherwise get invited to. Have kids who will be proud to call you their parent. And make your corner of the world meaningfully better. If anyone wants to seriously follow this recipe and just lacks the capital, happy to talk. - **Padday** (2025-06-12): https://x.com/Padday/status/1933116293122506981 - Hot, new startups get all the attention because they are raising money, getting new valuations, hitting milestones. Full respect to them, having done it, I know how hard it is. But I think what we've done at Intercom in the past two years is harder. We've built our best ever product in Fin, a brand new business growing at an incredible rate, and transformed the company for the AI age. All the details are here in Eoghan's post. Amazing effort from our entire team, very proud of them and us. Let me know if you've questions on what we've done and what we're doing! - **GavinNewsom** (2025-06-12): https://x.com/GavinNewsom/status/1933026025203794343 - https://t.co/s7RyFkKPSA - **TruebloodWesley** (2025-06-12): https://x.com/TruebloodWesley/status/1933008167388041546 - The NBA doesn’t have a Pacers/OKC problem They have a Stephen A Smith/ESPN problem - **GavinNewsom** (2025-06-12): https://x.com/GavinNewsom/status/1933006152784380095 - Pete sent 4,700 troops here (when they weren’t needed) without adequate fuel, food, water or a place to sleep. But don’t worry, he’s at a baseball game. https://t.co/2L1G9qN3LT https://t.co/C6g0327GCF - **GovPressOffice** (2025-06-12): https://x.com/GovPressOffice/status/1932976111866294699 - PETE, SEND THE MARINES ! https://t.co/30BU5OMj6M - **ibuildthecloud** (2025-06-11): https://x.com/ibuildthecloud/status/1932927634348138837 - I feel like the fear of websockets is unfounded these days. Rancher extensively used websockets and that was never a show stopper. Of course bidirectional communication is never easy and should be avoided. But websockets were fine. https://t.co/crT8IpD5hH - **signulll** (2025-06-11): https://x.com/signulll/status/1932926665971102119 - i was just talking to my friend (we were on a walk through madison square park) about how a ton of products today don’t struggle because they’re bad (they’re in fact really good). they struggle because no one can understand why they matter. what’s missing isn’t code or design. it’s clarity. it’s a story people remember & repeat. distribution should start with clear narrative. not ads. not virality. just a very simple reason someone gives a damn. - **destraynor** (2025-06-11): https://x.com/destraynor/status/1932919355529974240 - I love this @Fin_ai feature. It's both extremely powerful and quite hard to get it working really well, but we closely monitor our suggest acceptance rate and it's doing incredible numbers. https://t.co/jYJRHcWBHw - **sqs** (2025-06-11): https://x.com/sqs/status/1932912925343183340 - The best devs are devs+support+sales+PM+marketers. They win hearts and minds, build, and follow through. That feedback loop is so valuable. - **darraghcurran** (2025-06-11): https://x.com/darraghcurran/status/1932892944476107040 - @ciaran_lee been way more fun (and successful) with you back too!!! - **tanayj** (2025-06-11): https://x.com/tanayj/status/1932890832065212752 - Good overview of current pricing models of AI agents by @medinism @poyark https://t.co/vWZJA6GRNT - **TaliaGold** (2025-06-11): https://x.com/TaliaGold/status/1932883740117905592 - Intercom may be the most underhyped AI company. Customer support is transforming and their AI agent @Fin_ai is the best product on the market. I speculate Fin’s live ARR is the largest of the players, on top of a multi hundred million $ existing business. @eoghan returned to the business <3 years ago and is in full founder mode. - **destraynor** (2025-06-11): https://x.com/destraynor/status/1932881658966929598 - Aside from our product updates we’ve been pretty quiet about what’s going on at @intercom and @Fin_ai as a business. Until now. @eoghan shares some exciting highlights in this post, have a read👇 https://t.co/SSShb2tWdq - **mjackson** (2025-06-11): https://x.com/mjackson/status/1932873588601942052 - Don’t ever fall in love with the software you built. Someone will come along and make it mostly irrelevant before you know it. - **mjackson** (2025-06-11): https://x.com/mjackson/status/1932867013032226848 - RIP Brian Wilson, one of the greatest songwriters of all time. Such a legend. Good Vibrations, God Only Knows, California Girls, I Get Around, and so, so many more. Brian was an absolute gift to humanity. https://t.co/MPdXNo8dHr - **jshchnz** (2025-06-11): https://x.com/jshchnz/status/1932860659332743221 - Damn, really bummed to see that Brian Wilson passed Saw him with The Zombies in Santa Barbara back in 2019 & he was definitely struggling even then, but glad I ever got a chance to see such a legend Will be listening to Pet Sounds on repeat today in honor 🔁 RIP 🪦 https://t.co/mVQ2HzpHZY - **bgurley** (2025-06-11): https://x.com/bgurley/status/1932830327854043390 - I mentioned this on @BG2Pod. Enterprise customers need to declare if they support “open data” or “closed data.” And customers should move away from “closed data” vendors as fast as possible. https://t.co/qKgeV63zki - **garrytan** (2025-06-11): https://x.com/garrytan/status/1932827640131354947 - AI coding agents hit a wall when codebases get massive. Even with 2M token context windows, a 10M line codebase needs 100M tokens. The real bottleneck isn't just ingesting code - it's getting models to actually pay attention to all that context effectively. https://t.co/LQu6sq5EZk - **rohinlohe** (2025-06-11): https://x.com/rohinlohe/status/1932792001914941558 - once we start paying oems for mcp tool use, the great unbundling of saas will begin. it’s like a buffet with dishes from all your favorite restaurants — you pick your favorites and pay only for what you eat. you don’t have to get the fries at in n out just because you’re there. - **ketanhwr** (2025-06-11): https://x.com/ketanhwr/status/1932789071941652942 - Excited to share my first post on the Cloudflare Blog! https://t.co/IAfA35Neyb - **Cloudflare** (2025-06-11): https://x.com/Cloudflare/status/1932786155830829370 - We’ve recently added support for the FinalizationRegistry API in Cloudflare Workers. However, despite its availability, our general advice is: avoid using it directly in most scenarios. https://t.co/8NM6njSCZA - **thibodeau_96709** (2025-06-11): https://x.com/thibodeau_96709/status/1932770238291993004 - Thank you, New York. https://t.co/bKpZlBEgux - **dsp_** (2025-06-11): https://x.com/dsp_/status/1932708899129352416 - Good article https://t.co/1MXi7CawNX - **mjackson** (2025-06-11): https://x.com/mjackson/status/1932673739952726220 - React hooks are full of footguns that make it easy to introduce bugs in your code. I have trained hundreds of developers on React and how to use React hooks effectively. Here are a few of the most egregious and common ways people misuse hooks. 1/n - **frasergeorgew** (2025-06-11): https://x.com/frasergeorgew/status/1932665428691857656 - It’s taking over the world but in many ways Postgres is a shit database. High scale postgres deployments are incredibly brittle. You need staff to watch over them 24/7. One bad query can bring down the whole system. - **wunderacle** (2025-06-11): https://x.com/wunderacle/status/1932596903360803319 - It's been a while since I've seen a large enterprise doing so much for Open Source and how closely they've been working with open source maintainers to deliver value to their projects. A big heartfell thank you to @Cloudflare! Your support has been invaluable for @nodejs - **dkundel** (2025-06-10): https://x.com/dkundel/status/1932554359029879034 - The experience of using the OpenAI Agents SDK for all you Cloudflare Workers developers should be much better in the latest release. Right on time to make use of that 80% price cut of o3! Thank you @threepointone for the help! https://t.co/zgNYe2O2Fb https://t.co/SsZCHif3PV - **brian_scanlan** (2025-06-10): https://x.com/brian_scanlan/status/1932512208375722291 - Building great, reliable products that customers love using LLMs is not easy. Reminds me of using DNS and CDNs like 15 years ago - you have to use multiple providers. https://t.co/MKIx0L79yF - **miacmal** (2025-06-10): https://x.com/miacmal/status/1932455004192813105 - Looking to chat with a few folks about Cloudflare Secrets Store - if you've got feedback, use cases, or pain points, send 'em my way. - **nandafyi** (2025-06-10): https://x.com/nandafyi/status/1932452701163671611 - add a binding https://t.co/iyCZAZBlox - **GovPressOffice** (2025-06-10): https://x.com/GovPressOffice/status/1932447842683249033 - A ONCE GREAT AMERICAN CITY HAS BEEN OCCUPIED! https://t.co/R3EnZNIk55 - **pratik60** (2025-06-10): https://x.com/pratik60/status/1932442352154488977 - OpenAI is experiencing an outage since morning, but @Fin_ai is running smoothly! We use OpenAI for some components but have robust fallbacks in place for exactly this scenario. When you choose @intercom , you get this battle-tested reliability built in. - **yagiznizipli** (2025-06-10): https://x.com/yagiznizipli/status/1932431913509089281 - may @mikeal rest in piece. this is a devastating day for all of us at @nodejs. - **thdxr** (2025-06-10): https://x.com/thdxr/status/1932425952585220189 - i'm waiting for people to discover that those enterprise-y boilerplate heavy coding patterns boost LLM performance - **RedwoodJS** (2025-06-10): https://x.com/RedwoodJS/status/1932424616313815284 - RedwoodSDK & Client Side Routing in 15 lines: https://t.co/kXMZr1Ox9y - **RedwoodJS** (2025-06-10): https://x.com/RedwoodJS/status/1932376748999532696 - When we say you control every byte, we mean it. Today, we're going to show you an early preview of how to build your own single-page apps with client-side navigation. The contents are fetched on the server but diffed on the client. The result is buttery smooth, but absolutely no different mental models. You can use view transitions to make it feel like amaze. - **threepointone** (2025-06-10): https://x.com/threepointone/status/1932338654627647751 - rip @mikeal, a real one - **cramforce** (2025-06-10): https://x.com/cramforce/status/1932322373144895756 - RIP to a real legend of open-source and Node.js @mikeal As he would, he appropriately over-engineered his memory beyond death and solicited pull requests. Here is mine. While I'm not sure further PRs can be merged, I'm sure they are still appreciated https://t.co/H8tnJzFxaO - **brandinpodziem2** (2025-06-10): https://x.com/brandinpodziem2/status/1932290667327189230 - Kayla Thornton. Bucket. - **wesbos** (2025-06-10): https://x.com/wesbos/status/1932257297868189894 - @ljharb @mikeal Holy smokes - a true node OG. RIP brother - **ljharb** (2025-06-10): https://x.com/ljharb/status/1932254684171186631 - rip @mikeal, what a loss - **samlambert** (2025-06-09): https://x.com/samlambert/status/1932212418589184062 - The weather has been amazing lately so we decided to buy some E7 Campus APs and broadcast free and fast wifi across South Park so everyone can work outside. https://t.co/P8UDV7pfw4 - **nirgn975** (2025-06-09): https://x.com/nirgn975/status/1932154547553829105 - I've got to say, Cloudflare is winning me over. I came for the prices, but I'm sticking around for the awesomeness! Those new bindings visualizations? Simply top-notch! https://t.co/xCH8rjQKM7 - **mjackson** (2025-06-09): https://x.com/mjackson/status/1932101542297161928 - Async code in JavaScript is contagious. Any time you define an async function, the caller must also be async. async function fn() { // ... } async function mustAlsoBeAsync() { let result = await fn() // ... } 1/? - **davidhoang** (2025-06-09): https://x.com/davidhoang/status/1931916059910779263 - I don’t think it’ll be long until we see the concept of personal MCPs. Your personal web presence now has protocols that allow humans and agents interacting with your information based on permissions. MCP the new FTP. - **samlambert** (2025-06-08): https://x.com/samlambert/status/1931858156474519770 - i was at an event this week and 3 different people told me how good their experience of using @Fin_ai was. - **LegionHoops** (2025-06-08): https://x.com/LegionHoops/status/1931729223879643347 - “The way the game is digested by the fans sometimes. I think it’s a lot of boxscore watchers.” — Tyrese Haliburton (h/t @BrettSiegelNBA) https://t.co/w9WalYGRH2 - **gakonst** (2025-06-08): https://x.com/gakonst/status/1931701736655446168 - Passkeys will change the world. Passwords will be a thing of the past. Passkeys will be on every device and every passkey will have a stablecoin account. Compatibility issues across browsers / OSes will be fixed top down. We are building the best SDK for Passkeys. - **phuctm97** (2025-06-08): https://x.com/phuctm97/status/1931687557395513831 - Remote MCPs (with OAuth) are officially supported in Claude Desktop App, Claude Web App, Cursor, Windsurf, VSCode, OpenAI Responses API, Microsoft Copilot Studio. It’s time for remote MCP servers and MCP in general to get even more adoption! - **grinich** (2025-06-08): https://x.com/grinich/status/1931547930336628983 - rip bill atkinson ❤️ one of the greats (our 404 page is in homage to him and other early macintosh designers and engineers) https://t.co/xyJpB65iPE - **marcusnaughton** (2025-06-08): https://x.com/marcusnaughton/status/1931536027749061086 - @dok2001 @ngriffin_uk .well_known insanity - **jeff_weinstein** (2025-06-08): https://x.com/jeff_weinstein/status/1931522174508290210 - probably everything https://t.co/6tR4gQJdrN - **DhravyaShah** (2025-06-08): https://x.com/DhravyaShah/status/1931508776504557926 - this repository is probably worth a lot but who cares lol supermemory mcp is now open source https://t.co/tb3xhtQ9mG - **supermemory** (2025-06-08): https://x.com/supermemory/status/1931508487420588534 - We've open-sourced the supermemory MCP. Lots of goodies in this: - "Anonymous" users without auth - ad-hoc URLs for different connections - built on top of @Cloudflare durable objects oh, and using the supermemory api. link below https://t.co/FyKvyYBwfL https://t.co/BFQLAw27ps - **DhravyaShah** (2025-06-07): https://x.com/DhravyaShah/status/1931484491048169883 - supermemory MCP, which started off as just a small, random sideproject, built as a "customer" of the supermemory API, ...is now used by 63,000 people all around the world. DAILY. It might as well be the most used memory MCP in the market right now. How? Blog soon. https://t.co/3IlVLEztna - **eastdakota** (2025-06-07): https://x.com/eastdakota/status/1931481268845502604 - https://t.co/Er6SMwkviD - **mathemagic1an** (2025-06-07): https://x.com/mathemagic1an/status/1931449410074059062 - Multi-agent flows in Linear are so powerful. You can take on 5-10x more ambitious tasks. I just had it build out our marketing site in 20 minutes. I will likely make minor tweaks in Cursor and ship. This was unfathomable a year ago https://t.co/BMOhJogv0U - **rafalwilinski** (2025-06-07): https://x.com/rafalwilinski/status/1931371177345511693 - am I the last one on this platform to learn that you can use Claude Code via Github Actions and just vibecode on your phone using Github app??? https://t.co/GOqyhewq5T - **JoviDeC** (2025-06-07): https://x.com/JoviDeC/status/1931369577813877001 - Recently I've created three (LLM) applications fully on CloudFlare. I just love how every type of infrastructure is centrally on there, it's globally distributed, it has a generous free tier and no egress costs. In the near future they're even adding container deployments. - **CohenSite** (2025-06-07): https://x.com/CohenSite/status/1931193313841054012 - The LAX Metro Transit Center feels like a peek at a utopian vision of LA It’s clean, modern, and beautiful. Everything makes sense and works the way it’s supposed to. One day all of LA will be like this https://t.co/8c6fpRjUpU - **dok2001** (2025-06-07): https://x.com/dok2001/status/1931148686429684003 - Or just put Cloudflare in front. https://t.co/4jouGrqAoU - **imaurer** (2025-06-07): https://x.com/imaurer/status/1931141476919963668 - @kentcdodds We used @stytchauth for our open source MCP which will be available as a remote MCP soon: https://t.co/aW6RsHwMln - **kentcdodds** (2025-06-06): https://x.com/kentcdodds/status/1931138922869833972 - I'm starting to plan a workshop on remote MCP servers, and I'm planning on using Cloudflare. I'm pretty sure that I want to separate the concept of authentication from remote hosting, because I think authentication is a pretty big subject that would require a workshop all on its own. That said, assuming that learners already know about tools, prompts, and resources, I'm not sure how much to cover in a workshop about hosted MCP servers that doesn't involve authentication. What are some topics you think would be useful for a workshop like this to include? Again, assuming people have already gone through my MCP Fundamentals workshop where we build a local MCP server (so no need to have additional exercises on tools, prompts, and resources). I'm thinking exercises on 1. Configuring the Durable Object 2. Deploying to Cloudflare But that seems pretty weak. Maybe I should just lump this in with the authentication workshop 🤔 - **rakyll** (2025-06-06): https://x.com/rakyll/status/1931085121769390091 - In the last six months, we've been working on rebuilding our model serving infra by building an early version. We spent months coding, debating, hashing out everything -- before we write a design doc. Every major new system at Google is born this way. We are culturally back. - **syntaxfm** (2025-06-06): https://x.com/syntaxfm/status/1931080349884485951 - .@appfactory said it first @RedwoodJS https://t.co/T6fMeLTvrU - **yagiznizipli** (2025-06-06): https://x.com/yagiznizipli/status/1931056836738011647 - In the past 48 hours I've written 10k lines. I need a long holiday. - **dok2001** (2025-06-06): https://x.com/dok2001/status/1931043270030700682 - @backpinelabs Agree. Though I don’t even think that is one of the biggest things to come this summer. - **ritakozlov** (2025-06-06): https://x.com/ritakozlov/status/1931033008611397760 - today marks 9 years for me at @cloudflare! when i first joined, i was deploying lua code as part of our weekly edge releases to help customers customize cloudflare for them. today we have a full developer platform that spans compute, storage, databases, AI… 🤯 we’ve come such a long way, and every day still feels like we’re just getting started! i’ve never been more excited about the future, and i couldn’t ask for better people to get to ship alongside 🧡 - **tadasayy** (2025-06-06): https://x.com/tadasayy/status/1930993346824241518 - (1/7) Inspired largely by @_catwu's post claiming that Claude Code + Opus 4 was a new unlock for software engineering, I dug in to try it out myself. It sure delivers. The secret sauce: combine Claude Code with git worktrees 🧵 https://t.co/W6SkaqvceS - **SportsCenter** (2025-06-06): https://x.com/SportsCenter/status/1930844912594952283 - Teams were 0-182 in the NBA Finals entering Thursday in this scenario 😮 Pacers are different. https://t.co/6Hi4KALgxY - **samlambert** (2025-06-06): https://x.com/samlambert/status/1930842973060973035 - sharded metal go brrrrrr. this is a single database doing 130GB/s of filesytem throughput. that's ~7 million write IOPs. https://t.co/ou62eIg0OR - **hunvreus** (2025-06-06): https://x.com/hunvreus/status/1930840247144943943 - It's time to move @pagescms from Next.js to React Router + @honojs. If I needed SSR I'd probably stick with it, but here I just need a regular SPA + API. I'm also interested to see if I can go back to @Cloudflare Workers + D1. - **NBA** (2025-06-06): https://x.com/NBA/status/1930837234389070281 - SOUND UP 🔊🔊🔊 The Korean SPO TV live call of Tyrese Haliburton's Game 1 winner is UNREAL 🔥 https://t.co/bDs05zeEon https://t.co/heI0ELIQlu - **ramonashelburne** (2025-06-06): https://x.com/ramonashelburne/status/1930825310934548908 - I was here for Aaron Gordon’s buzzer beater in Game 1 of the second round series and I can tell you this building is a hell of a lot quieter after what Tyrese Haliburton and the Pacers just did tonight. Though maybe not as quiet as the Garden after that Game 1. These playoffs have really been incredible - **Rachel__Nichols** (2025-06-06): https://x.com/Rachel__Nichols/status/1930824336065094109 - Sports are fun. - **CaitlinClark22** (2025-06-06): https://x.com/CaitlinClark22/status/1930823209487606168 - YOU CANT MAKE IT UP 😂😂😂😂😂😂😂😂😂😂😂😂😂😂😂😂😂 - **benallfree** (2025-06-06): https://x.com/benallfree/status/1930822752761765908 - ~~dofs@0.0.2 released~~ Durable Objects File System (dofs) the native virtual filesystem for @Cloudflare Durable Objects, just got easier. I added a `withDofs()` helper to simulate a trait helper. https://t.co/ReuoxVixX5 - **ritakozlov** (2025-06-06): https://x.com/ritakozlov/status/1930788611303293139 - “ai transformation” is going to be the “digital transformation” of the next 20 years. calling it now. - **sama** (2025-06-06): https://x.com/sama/status/1930785056194539779 - we have been thinking recently about the need for something like "AI privilege"; this really accelerates the need to have the conversation. imo talking to an AI should be like talking to a lawyer or a doctor. i hope society will figure this out soon. - **sama** (2025-06-06): https://x.com/sama/status/1930785054005076100 - recently the NYT asked a court to force us to not delete any user chats. we think this was an inappropriate request that sets a bad precedent. we are appealing the decision. we will fight any demand that compromises our users' privacy; this is a core principle. - **davidaldridgedc** (2025-06-05): https://x.com/davidaldridgedc/status/1930768213593469289 - Adam Silver, in his opening remarks, in discussing the issue of market size going into these Finals, cites data from Oklahoma City Mayor David Holt: About 300 million Americans, or 88 percent of the populace, live in cities smaller than Oklahoma City and Indianapolis. - **Rachel__Nichols** (2025-06-05): https://x.com/Rachel__Nichols/status/1930763961009369155 - Rick Carlisle opens his press conference by expressing dismay over yesterday's report on Doris Burke's job security as part of ESPN's lead broadcast team: "She has changed the game for women in broadcasting...Doris is a great example of courage and putting herself out there. It's so sad to see these reports leaked before such a celebrated event." - **CloudflareDev** (2025-06-05): https://x.com/CloudflareDev/status/1930756590887661831 - Some highlights from our VP developers + ai @ritakozlov's talk at @aidotengineer #aiewf: Building Agents (the hard parts!)🧡🚀 https://t.co/AUOfb1QeDn - **CloudflareDev** (2025-06-05): https://x.com/CloudflareDev/status/1930724892707901556 - Miss our demo w/ @WorkOS? Here's the agenda https://t.co/nEUYB8ZAS8 And catch our VP of devs + ai @ritakozlov @ 2:40PM in Golden Gate ballroom C talking about the hard parts of building agents! @aiDotEngineer #aiewf https://t.co/HCRqZRqrEP - **mwylde** (2025-06-05): https://x.com/mwylde/status/1930717339777179823 - Another @ycombinator company joining @CloudflareDev! https://t.co/ZHJko2YXmZ - **threepointone** (2025-06-05): https://x.com/threepointone/status/1930675651578601587 - Hello from the Cloudflare ETI India office :) https://t.co/oHzgcsPXg2 https://t.co/Xs8P71Rhoa - **dinasaur_404** (2025-06-05): https://x.com/dinasaur_404/status/1930669836134445195 - Nice to see Cursor support remote MCP — both auth & streamable HTTP! Hope to see more MCP clients follow their lead :) - **KankaniAbhishek** (2025-06-05): https://x.com/KankaniAbhishek/status/1930656077382824248 - Hey there @Cloudflare 👋 https://t.co/EB11Eu0g6I - **mwylde** (2025-06-05): https://x.com/mwylde/status/1930649163911622975 - Excited to talk about how @ArroyoSystems does streaming incremental view maintenance with DataFusion! https://t.co/KAzpbmzuts - **brian_scanlan** (2025-06-05): https://x.com/brian_scanlan/status/1930636345937920010 - I've referenced this tweet dozens of times when pitching Intercom as a place to work. https://t.co/8nqbutfoiL - **appfactory** (2025-06-05): https://x.com/appfactory/status/1930634397650145661 - I built a @RedwoodJS cloud development environment running in @CloudflareDev's containers and workers. Short demo, code coming soon. https://t.co/2oHpbFeXiV - **benallfree** (2025-06-05): https://x.com/benallfree/status/1930633764226707765 - yesterday's coding ended early because i took a day trip to SF and got to meet some of the @Cloudflare crew. i met @dok2001 @oroup @ritakozlov @dinasaur_404 @irvinebroque and a few others i haven't tracked down on x yet. (if you're the guy who recognized my @PinkGorillaLLC shirt hmu) from the first interaction, i was hit with an `oh shit` (in a good way) feeling. these are highly competent people. when you meet someone with high competency, there is an ease and friendliness. a sense that long explanations, or even short explanations, are not necessary because they already know. words feel slow compared to the mind meld already happening. i went to cfhq because we have been exploring the idea of me joining cf in some capacity. if i did, i would be in a better position to develop projects like AdminDO and dofs, and persistent storage for cf containers. there are pieces of these projects that really want to live in the underlying infrastructure. i'd have more visibility and influence if i were on the inside. from my point of view, i'm on a trajectory to develop the same kind of infrastructure anyway, so if i can multiply the impact of my work by accomplishing it through cf then all the better. having the support of competent people and being able to support them too sounds really good. so i hope something happens! and you're talkin to a guy who is extremely averse to structure and authority not backed by competency. i also really love teaching, sharing, tool making, and spiking new ideas, building infra, ts, rust, hardware, asm.... i guess i like software at every level. building in public. trying to do something, hitting walls, taking sidequests, and making the dx better for the next person. i often feel that my top level projects are actually just conversation starters with the universe, meant to reveal the real work. for example, AdminDO has already yielded vite-plugin-unpkg which is a humble but important contribution to the no-build movement that seems to be emerging. frontend tooling and frameworks have become so complex and unwieldy, i see a landscape ripe for contraction back to simplicity. it's why i like @pocketbase, why i made https://t.co/1Eq3tC7z5C and https://t.co/ZnfBGT70x5 and why projects like @htmx_org and data-star @DelaneyGillilan seem vital. cf plugs into this simplicity too. they've abstracted away almost all the complexity of shipping scale to zero. so yeah, i think there is something going on at the intersection of all this and it was fun to meet some of the ppl involved thanks for the generosity! - **benallfree** (2025-06-05): https://x.com/benallfree/status/1930633765753413818 - onward to today's goals. we left off yesterday with the about plugin able to list the other installed plugins. i actually had to check the git log to remember that. next, there is something bothering me about the admindo default route. it has these hard-coded plugins, but now that the plugin names are the package names, we can make this dynamic. - **legionsdev** (2025-06-05): https://x.com/legionsdev/status/1930618393117478945 - Invoicely is now part of Cloudflare's OSS program — Project Alexandria! 🚀 Soon migrating invoicely project to Cloudflare — big things ahead. Big features coming! Keep invoicing! @CloudflareDev https://t.co/hWado5eopy - **yacineMTB** (2025-06-05): https://x.com/yacineMTB/status/1930616329788166513 - You should be a generalist. Generalists can make tradeoffs across the entire stack. It's better in every way, trust me. You can do things that other people couldn't have, you can be creative. The only cost to do so is your ego. You need enough humility to be a permanent amateur - **yusukebe** (2025-06-05): https://x.com/yusukebe/status/1930580453024526581 - I met the Hono friend @MathurAditya7 again https://t.co/bRX2t8nHFT - **asleMammadam** (2025-06-05): https://x.com/asleMammadam/status/1930533081355800682 - it's here y'all! happy to announce now @CloudflareDev is backing https://t.co/8rAeBOeEF6's data infrastructure. thank you all for supporting the project and making this possible! it's a dream coming true. https://t.co/DBTwyRgX0p - **craigsdennis** (2025-06-05): https://x.com/craigsdennis/status/1930469315339563361 - @boristane https://t.co/eVlivbmG5I_SINK - **yagiznizipli** (2025-06-05): https://x.com/yagiznizipli/status/1930421529801593305 - I realized I am most productive when I work on something that requires 3 languages (c++, rust, typescript). All have different mental models that help me focus more. Don’t know why… - **TobinSouth** (2025-06-04): https://x.com/TobinSouth/status/1930411129827930501 - Unreal demand for watching a random Aussie explain vibing MCP into the enterprise at @aiDotEngineer https://t.co/gSJwlkimYe - **benallfree** (2025-06-04): https://x.com/benallfree/status/1930398690629562686 - What did y’all do today? https://t.co/Qi5lBMceVs - **G4brym** (2025-06-04): https://x.com/G4brym/status/1930382554370097581 - Cloudflare's new Worker Templates lets you deploy R2-Explorer in just a few clicks 🚀 https://t.co/eS47sSJ9qf - **stackblitz** (2025-06-04): https://x.com/stackblitz/status/1930380832641818752 - As we move into onboarding more OSS projects to https://t.co/5AUOsdVT8j and are releasing the next million preview packages, we are thrilled to announce that @Cloudflare is now backing our data infrastructure! https://t.co/5VT777gKDO https://t.co/T2kMCgf24z - **kentcdodds** (2025-06-04): https://x.com/kentcdodds/status/1930368331611287908 - I love how @zeeg found out he was speaking two days ago and comes out with one of the best talks because he has real world experience building with MCP and can talk about the issues with MCP that are being worked on. https://t.co/WfYNZExdQq - **fnthawar** (2025-06-04): https://x.com/fnthawar/status/1930367595670274058 - I'm seeing a concerning trend in talking to CTOs/CEOs about the cost of tokens i.e. Can I afford an extra $1-10k/month per engineer as they use @cursor_ai / @windsurf_ai / @GitHubCopilot etc? So they clamp down on spending 😬 1. Do not use the default / cheapest models. Yes all the models are getting better, but there is a big difference between the default Cursor model, the small @GoogleDeepMind Gemini Flash and @OpenAI O1 Pro. Splurge for a better model. 2. You might use completely different models for planning and actual coding. Exampe — O1 pro for planning / requirements and then @AnthropicAI Claude 4.0 for coding 3. If your engineers are spending $1k/month MORE because of LLMs and they are 10% more productive, then that is TOO CHEAP. i.e. Anyone would KILL for a 10% increase in productivity for only $1k/month. Totally worth it 4. If you are worried about engineers spending $10k/month on tokens -> I wouldn't worry. If they are spending this much, DM me because I want to know what exciting things they have conjured up to spin up LLMs on. If they are spending this much AND getting value out of it, you are winning (DM me!) 5. Short of tempting Goodharts Law, you could reward your engineers (well, all employees!) for maximal AI spending -> this means they have figured out how to leverage the LLM well 6. The workflow is moving quickly from AI autocomplete to "I am the architect of a team of agents". This means a single person can wield of a team of LLMs to build their code. Makes sense to see your corresponding token bill go 📈 7. Worry about LOC (lines of code), as the current cohort of models is very code verbose. This will hurt you later if you're building for production vs. just prototyping. It will take smart prompting / rules / iterations to build the short / sweet elegant code you actually want 8. My Cursor deployment quickly increased when non-RnD folks started using it to build their workflows. Folks in Support, Commercial, Ops and more can build things now 👊🏽. The last thing I want is to stop these licenses from growing. I know folks @ShopifyEng who were proud to show up on the Top 10 token spend leaderboard due to the valuable work being done - **EastlondonDev** (2025-06-04): https://x.com/EastlondonDev/status/1930367379848020203 - This release is absolutely huge. Cursor have cooked tons of new functionality. Code review agent Easy MCP install Background Agents in the cloud Memories and more and more stuff https://t.co/khkKCEQqYX - **kentcdodds** (2025-06-04): https://x.com/kentcdodds/status/1930367327717208548 - I'm so excited to see @digitarald announce on stage that @code insiders fully supports the MCP spec (including sampling!!!). Things are moving! https://t.co/Pbt0hFXqyz - **kentcdodds** (2025-06-04): https://x.com/kentcdodds/status/1930365300014493789 - Just sold out the 5th MCP workshop before the discount period was over. Unbelievable interest in this stuff. - **KentonVarda** (2025-06-04): https://x.com/KentonVarda/status/1930364281381630200 - A lot of people seem to be looking at my Claude Code prompts to try to learn some sort of techniques. Folks, just to be clear, I didn't know what I was doing, I just talked to it like I'd talk to a human. - **iannuttall** (2025-06-04): https://x.com/iannuttall/status/1930361854884491729 - Btw that Cursor 1.0 post I just published was done by an AI agent I vibe coded in Cursor: - Cloudflare Worker monitors changelog every minute - Extracts page content to Markdown - Stores to Cloudflare KV if doesn’t already exist - If KV exists, check the diff of the two - Pass diff to Gemini 2.5 Flash - Gemini writes the recap - Pings me on Telegram - If I confirm, publishes to X You can just build things. - **msfeldstein** (2025-06-04): https://x.com/msfeldstein/status/1930361446719992102 - Lots of MCP goodies in here - one click install - oauth - headers support - admin controls - better debugging and some other goodies https://t.co/smBggFw6uU - **dsp_** (2025-06-04): https://x.com/dsp_/status/1930331762372284718 - Okay I don’t speak Turkish, but that Turkish Airline has their own MCP server is really cool. https://t.co/AtdtcVBHXr - **dinasaur_404** (2025-06-04): https://x.com/dinasaur_404/status/1930329490439450889 - The secret sauce to building an MCP server that doesn't suck? 🧠Memory🧠 Let your server remember the user’s past tool calls/preferences so every interaction can get better! If you want to know how I built GoodReads into an MCP server in 30 minutes on @Cloudflare, keep reading! https://t.co/ebXC31LtgP - **shcallaway** (2025-06-04): https://x.com/shcallaway/status/1930327724415168702 - Using MCP to feed real, live goats. Unexpexted things happen at @aiDotEngineer. Credit to @TobinSouth for this great talk on creating secure, production-ready MCP servers! https://t.co/M8hLxh1R0o - **rrnn** (2025-06-04): https://x.com/rrnn/status/1930316276515172627 - Client side packet loss metrics from a customer at scale moving from Twilio to Cloudflare TURN https://t.co/y9qstPQsPj - **boristane** (2025-06-04): https://x.com/boristane/status/1930302487602573790 - chat am I doing this right? https://t.co/iiceGqb1hC https://t.co/qCrmy1m1oa - **kentcdodds** (2025-06-04): https://x.com/kentcdodds/status/1930293591190188236 - My talk at @mcpsummit has been published! I talk about what stands between us and an Ironman's Jarvis-like experience. I'm trying to nerd snipe you! Please Make Jarvis So I Don't Have To https://t.co/ntgu3GBNpU - **threepointone** (2025-06-04): https://x.com/threepointone/status/1930283960216547409 - the more I work on this, the more it feels like a ☯︎ yin yang situation; openai agents give the abstractions to build stuff, and cloudflare agents are a great compute + storage container for it to really shine. banger. case in point: their human-in-the-loop abstraction works so well with using the cf-agent storage / sync capabilities, meaning you can build and deploy real apps so quickly. example code landing tomorrow after I clean up some stuff. - **kyliestew** (2025-06-04): https://x.com/kyliestew/status/1930283127274123285 - we're just getting started :) https://t.co/RkQRlr0ZiZ - **boristane** (2025-06-04): https://x.com/boristane/status/1930280115386364286 - there's so much terrible software in the world - **_tanushreeeee** (2025-06-04): https://x.com/_tanushreeeee/status/1930277204258369753 - was fun jamming with @rohinlohe on evals for a new Workers feature! if you’re building new ai products/features and want to learn about evals or validate your approach, reach out — happy to chat https://t.co/KkWzYUOGTR - **appfactory** (2025-06-04): https://x.com/appfactory/status/1930273114597781740 - I'm building a cloud-based text editor on top of Cloudflare's containers. The container runs a HTTP server that allows me to modify files inside of it. My editor, which is @RedwoodJS, interacts with this node-based process in order to modify the file system in the container. I cannot tell you how many times I've wished that I could just use RedwoodSDK instead of this node HTTP interface. I do wish workers could access the file system. It would make my life so much easier. And I get why they can't. It's not a valid use case! - **kuri_tter** (2025-06-04): https://x.com/kuri_tter/status/1930256669226700892 - 昨日のMCP開発体験の記録感想日記を書きました！ microCMSブログ記事データとD1の体重データに接続するMCPを作ってみました。めっちゃ楽しかったです！ Cloudflare Workers Workshop in Tokyo でMCP開発体験した https://t.co/ozyvo2bJf5 #workers_tech - **grinich** (2025-06-04): https://x.com/grinich/status/1930249176203305446 - Are you building an MCP server? WorkOS @AuthKit now natively supports MCP auth! We think it's the easiest way to add identity and authorization to your project. Check it out: https://t.co/h00YvribQr - **iannuttall** (2025-06-04): https://x.com/iannuttall/status/1930239263867589039 - @malpern https://t.co/QWOnSUTNUK - **jamesdevonport** (2025-06-04): https://x.com/jamesdevonport/status/1930238022966005809 - Workers continues to get better and better. I love how they are spending time building a lovely web UI where you can update everything visually. https://t.co/7oogr2jC5b - **CloudflareDev** (2025-06-04): https://x.com/CloudflareDev/status/1930229797533155503 - JUST SHIPPED: you can now visualize all of the services your Worker is connected to directly in the dashboard — and connect it to more with a click or two! ⛅️ Databases. KV stores. Other Workers via RPC. R2 buckets. All there in a new canvas. https://t.co/zryzF1EKth - **RozenMD** (2025-06-04): https://x.com/RozenMD/status/1930216035925303489 - loving the new Bindings view in the Cloudflare Workers dash https://t.co/DeUHhXmKh8 - **iannuttall** (2025-06-04): https://x.com/iannuttall/status/1930207655941075367 - Hot take 🌶️: using Cursor to vibe code a Python script hosted on Cloudflare Workers is 100x easier than trying to build an n8n workflow - **threepointone** (2025-06-04): https://x.com/threepointone/status/1930201116974657678 - needed a couple of hacks, but we have liftoff. cloudflare ⨉ openai agents https://t.co/BEQAbsqZcD https://t.co/tUoPetIZ3X - **_pion** (2025-06-04): https://x.com/_pion/status/1930127272197214615 - FFMPEG MERGED WEBRTC SUPPORT!!! https://t.co/rvTbGwQImh The new things that are about to be built is gonna be huge. I am so excited. Sub-second video on a open protocol it’s a great future. For a what/why on WHIP check out the broadcast box README https://t.co/08yULbpwhi - **boristane** (2025-06-04): https://x.com/boristane/status/1930083497571692779 - I have a new theory about evals - **OaklandFireCA** (2025-06-04): https://x.com/OaklandFireCA/status/1930074662207373450 - Update 4) Battalion Chief Anthony Sanders provides an incident summary from the Skyline Blvd fire this evening, and select images of firefighters bringing the rescued dogs out of the home for additional care. https://t.co/XB7OTfD10K https://t.co/2E1ralWTda - **CloudflareDev** (2025-06-04): https://x.com/CloudflareDev/status/1930058368720482352 - 🧡🚀 https://t.co/uGNCGBvLZg - **shotbysteph30** (2025-06-04): https://x.com/shotbysteph30/status/1930051978404737421 - Steph asking for his ring in the middle of the 3rd quarter on the road in an elimination game is hands down the greatest moment in NBA History https://t.co/NR6nBodtHS https://t.co/nDrRYcMrKr - **mehrantsi** (2025-06-03): https://x.com/mehrantsi/status/1930026774093496692 - @elithrar I built this using agents sdk a while back and had a blast 🙌🏼 https://t.co/lsEYWS5TiG - **rafalwilinski** (2025-06-03): https://x.com/rafalwilinski/status/1929964513211588993 - Guess the office https://t.co/sp3TOxVWdl - **ryan_sherlock** (2025-06-03): https://x.com/ryan_sherlock/status/1929957724441223583 - Watching internal demos of this @intercom (and now live!) has been so cool. Opening up phone support to a whole class of customers that couldn't previously consider it. https://t.co/d1DRr1Mrtu - **antonosika** (2025-06-03): https://x.com/antonosika/status/1929937351947461040 - shoutout: @Cloudflare for reliable hosting @FabianHedin who checked our metrics before we went above 1337 https://t.co/6GKAuveIcM - **zaygranet** (2025-06-03): https://x.com/zaygranet/status/1929933369581941265 - if we could either: - Increase SF temperature by 10 degrees - Keep restaurants open until 11 This city would be paradise - **marbemac** (2025-06-03): https://x.com/marbemac/status/1929922350549479564 - Type-safe resumable streams and multi-user chat, courtesy of @Cloudflare durable objects, @zero__ms, and @orpc - how does it work? Streaming every token through pg / zero is not a great option at scale, so we're using SSE and cloudflare durable objects to handle the token streaming, and pg / zero for the durable persistence once streaming is complete. 1. to kick off generating a reply, we insert a new message into postgres with status "streaming" and empty content 2. this new empty message is sync'd to the client via Zero 3. on the client, when rendering any message that has status "streaming", open an SSE connection via ORPC to a durable object instance representing the chat thread 4. we're opening an SSE connection per "streaming" message, not per thread. multiple messages can be streaming at the same time, within the same thread 5. token stream parts are kept in memory in the durable object, and streamed to connected clients over SSE 6. if a new client connects (or old client re-connects), it is immediately caught up to the present by streaming down the in memory parts on connect 7. on llm stream complete, update the message in pg with the final message parts, set status to "completed", and close sse connections 8. the message part and status change is sync'd to client via zero - client now uses those parts to render the message instead of the previously streamed parts Thus there are only two postgres operations per message - initial insert, update on streaming complete. This should hopefully scale to the moon. - **kentcdodds** (2025-06-03): https://x.com/kentcdodds/status/1929912345284796882 - My @aiDotEngineer online talk has been published! In this talk, I explain why MCP has exploded in popularity recently and how it will impact you as a web developer. Check it out at the link below! https://t.co/RDOXY4sb5m - **elithrar** (2025-06-03): https://x.com/elithrar/status/1929911776251977838 - Great seeing more and more teams build agents + MCP servers using the Agents SDK 😎 https://t.co/8nRrOyXsN2 - **CloudflareDev** (2025-06-03): https://x.com/CloudflareDev/status/1929898337110536328 - Want to learn how @knocklabs is using the Agents SDK to build agents with human-in-the-loop capabilities? Read on! https://t.co/mmweiUzTtI - **Cloudflare** (2025-06-03): https://x.com/Cloudflare/status/1929886768880296431 - In this blog post, we'll use Knock (@knocklabs) and the Cloudflare Agents SDK to build an AI Agent for a virtual card issuing workflow that requires human approval when a new card is requested. https://t.co/BSdj8bdGjU - **ritakozlov** (2025-06-03): https://x.com/ritakozlov/status/1929884485522202686 - omw to sf for @aiDotEngineer! making a point this year of actually going to the conferences that i was feeling fomo about in previous years and not regretting it one bit please say hi if you’re there! 👋👋👋 s/o and congrats @swyx for making this happen - **shreygups** (2025-06-03): https://x.com/shreygups/status/1929862800064893290 - i've cranked out 3 super in-depth prompts for an overhaul we're doing for our ai generation pipeline and i truly have to thank my k-12 teachers for this. it's really paying off right now. i wouldn't be able to write this fast without them - **__syumai** (2025-06-03): https://x.com/__syumai/status/1929857868192506230 - CloudflareのGlenさんのmcp-remote、めちゃめちゃ便利で助かってる #workers_tech https://t.co/eq50OxjgGX https://t.co/5adD5EUCBe - **zebassembly** (2025-06-03): https://x.com/zebassembly/status/1929708557194482128 - Swift looks like such an interesting language for the backend, it saddens me so much that it was trapped in the Apple ecosystem while Go became so commonplace. https://t.co/4msVNO6aMx - **samgoodwin89** (2025-06-03): https://x.com/samgoodwin89/status/1929700808066846988 - Alchemy's tests now run in 100s instead of 1-1.5 hours. Fast take-off is imminent 🚀 https://t.co/HKEDEgBYYV - **mitchellh** (2025-06-03): https://x.com/mitchellh/status/1929695398354211296 - At this point most of the programmers I respect the most are AI positive and finding a lot of value in it. And of course understanding the limits, as with any good tool. @tqbf, Kenton Varda, @antirez, and many more. This post is mostly perfect: https://t.co/0tVXaqeMEs - **_edmundhung** (2025-06-02): https://x.com/_edmundhung/status/1929663928503615666 - In the next version of the @Cloudflare Vite plugin, you can run multiple Workers in separate processes with service bindings between them—and it just works. Stoked to see it finally ship. https://t.co/0LAiY5R6aa - **thdxr** (2025-06-02): https://x.com/thdxr/status/1929652559460889032 - it is very hard to ship fewer features - **_ashleypeacock** (2025-06-02): https://x.com/_ashleypeacock/status/1929615041096978939 - If anyone wants to build something like this, Cloudflare’s Email Workers are the perfect tool for the job. You can effectively invoke code whenever an email is received, and it’s incredibly simple to setup. https://t.co/2Vf6YcIqpg - **appfactory** (2025-06-02): https://x.com/appfactory/status/1929613931792028129 - Lots of people describe RedwoodSDK as "invisible", which is ironic because we try and give you access to every single part that affects the request and response. Our goal is to make you only have to care about the software you want to write, rather than the framework or the platform that it runs on. - **ryanflorence** (2025-06-02): https://x.com/ryanflorence/status/1929598696951632033 - https://t.co/FhoS2bFzMj - **ryanflorence** (2025-06-02): https://x.com/ryanflorence/status/1929598360723640371 - compose server/client content infinitely? style your app immediately, at runtime, with no extra tooling? cache views in the browser for fast back-forward navigation and on the server for faster everything? - **ryanflorence** (2025-06-02): https://x.com/ryanflorence/status/1929597856190476784 - get llms to generate partial views of your app on the fly expose an mcp server of your actions and data with the same abstractions as building the UI? embed a chat in your app that can help the user navigate around and perform actions? replace your whole dnag UI w/ the chat? - **ryanflorence** (2025-06-02): https://x.com/ryanflorence/status/1929597176566763920 - load data in the server and client with the same APIs? optimize network calls by only fetching blocks of UI you need for navigation and action revalidation? run your server in the browser for local first + sync? - **ryanflorence** (2025-06-02): https://x.com/ryanflorence/status/1929596345893244986 - Remix 3 is going to be so much more than just the current bundler conversation going on What if you could easily manage focus? cleanup actions instead of queuing them up? build composable custom events? them a batteries-included ui library? start fast w/ a built-in data store? - **kentcdodds** (2025-06-02): https://x.com/kentcdodds/status/1929561533820145824 - "Vite is now distributed as ESM only" This is good. Everything should be this way. https://t.co/rTsxV4ATjH - **BraydenWilmoth** (2025-06-02): https://x.com/BraydenWilmoth/status/1929550068740505620 - I think I've cracked the code on making Durable Objects data browsable in the dashboard... and also let you view all the instance ID's you've used... and let you get the DO's identifier from alarms... and more... ...now back to work - **dsp_** (2025-06-02): https://x.com/dsp_/status/1929532379472785550 - I actually want to interact with my coding agent via email. Send a mail with instructions, get a thread with results and PR link back. - **danprimack** (2025-06-02): https://x.com/danprimack/status/1929488523188420665 - Chime launches IPO roadshow. $9.1b market cap if it prices in the middle, vs. the $25b private market valuation. VC-backed unicorns maybe are finally done clinging to their ZIRP-era valuations. Seeing this more and more. - **whoiskatrin** (2025-06-02): https://x.com/whoiskatrin/status/1929453183064998204 - GUESS WHAT… I’m joining @Cloudflare to work on Workers' AI and agents thrilled to be part of this team that’s shaping the internet - fast, secure, and open. i’ll be working with incredibly smart people on products i truly believe in. can’t wait to start building ☁️🧡 - **samgoodwin89** (2025-06-02): https://x.com/samgoodwin89/status/1929413720884855195 - I got all the excited by fanning out Devin and getting working code. But quickly ran into the wall of slow tests. Iteration speed is your most important metric. How quickly can you verify the change? - **martin_casado** (2025-06-02): https://x.com/martin_casado/status/1929390376185405743 - Programming with AI is what you thought programming would be like prior to learning it. - **signulll** (2025-06-02): https://x.com/signulll/status/1929389031013417257 - the “sharing” in the so called sharing economy was a trojan horse. people say they want community, but they pay for control, privacy, & predictability. airbnb morphed from air mattress in someone’s loft to suburban mcmansion getaways. the idea of an uber is on the same path. people don’t want to chat with a stranger or sit in their smells… they want clean, silent, autonomous teleportation. so question is what’s keeping uber afloat in the markets? - **liadyosef** (2025-06-01): https://x.com/liadyosef/status/1929154529250971789 - Plans changed - I won’t make it to the (truly awesome) @aiDotEngineer summit in SF next week. I’m giving away my ticket *for free*. If you’re an early-stage solopreneur or just starting in AI and want to attend one of the best AI conferences in the world - DM me. - **OaklandFireCA** (2025-06-01): https://x.com/OaklandFireCA/status/1929054913058759016 - Update 3) Fire Chief Damon Covington provides incident update at the scene of tonight‘s 3 alarm vegetation fire near Grass Valley and Skyline Boulevard. Fire is now contained forward progress stopped. Crews will be on scene and in the area throughout the night. https://t.co/QzsRolj0JG https://t.co/h5QjvTNhIx - **OaklandFireCA** (2025-06-01): https://x.com/OaklandFireCA/status/1929044112260383027 - Update 2) As of 10:10pm, the Forward Progress of this fire is stopped. Approximately 60 firefighters on scene, and crews have fire surrounded. https://t.co/MjXaZW5lbS - **OaklandFireCA** (2025-06-01): https://x.com/OaklandFireCA/status/1929040241282707774 - Update 1) 57 mins into this incident. OFD, Cal Fire, East Bay Regional Park District, Oakland Police on scene. No structures threatened at this time. Incident command post is at Skyline and Grass Valley. Media should contact OFD PIO with questions — 5102899822. https://t.co/sywJ2IOIY9 - **OaklandFireCA** (2025-06-01): https://x.com/OaklandFireCA/status/1929032472395731372 - A vegetation fire in the area of Skyline and Grass Valley has gone to a 3rd alarm. Approximately 50+ firefighters responding to this incident. No structures are currently threatened. - **BleacherReport** (2025-06-01): https://x.com/BleacherReport/status/1929022040591229270 - ERNIE AND THE CREW SIGN OFF OF INSIDE ONE LAST TIME ON TNT 🥹🙌 https://t.co/CCcrg2V4t1 - **thdxr** (2025-05-31): https://x.com/thdxr/status/1928949794216751446 - we're building a share page for opencode to snapshot exactly what the AI does we're also working on a new set of benchmarks to track how these tools perform on a real world codebases this share page will snapshot every run so we can qualitatively review performance https://t.co/TaoRa44zbv - **prdonahue** (2025-05-31): https://x.com/prdonahue/status/1928924459659829536 - the most san francisco shirt i’ve seen for sale https://t.co/fX9uDgPI11 - **benallfree** (2025-05-31): https://x.com/benallfree/status/1928901882396741815 - ~~accouncement~~ kysely-do@0.0.1-rc.0, a Kysely driver for @Cloudflare Durable Objects, has been released. why this matters: now we are 1 step closer to using @better_auth with Durable Objects - **patio11** (2025-05-31): https://x.com/patio11/status/1928878728760930759 - *blinks* https://t.co/ZZmVGm7sxg - **samgoodwin89** (2025-05-31): https://x.com/samgoodwin89/status/1928794521422164153 - Ok this is actually getting crazy ... Devin just cloned the Stripe Terraform provider (10 new resources covering Stripe's API. I am going to apply this to all Terraform providers soon enough! Breadth. Is. No. Moat. https://t.co/zAjPlhjiFj https://t.co/HoPTt7G813 - **samgoodwin89** (2025-05-31): https://x.com/samgoodwin89/status/1928745351151857739 - The Miniflare codebase over uses Zod to the extreme. Please don't do this. https://t.co/wOEGliFPQM - **martin_casado** (2025-05-31): https://x.com/martin_casado/status/1928643589011013954 - SF!!!! ✌️🫡 https://t.co/KHUcXou5jS - **ekzhang1** (2025-05-30): https://x.com/ekzhang1/status/1928593804963258724 - I have made my pilgrimage to the holy ground of cloud systems research 🫡 https://t.co/YvwWXDtj3D - **eastdakota** (2025-05-30): https://x.com/eastdakota/status/1928585201602474074 - @levie @dok2001 @ashleymayer Interesting to think through how we may be in a unique position to do something clever here. - **eastdakota** (2025-05-30): https://x.com/eastdakota/status/1928583442272276802 - @levie I have an idea. Cc: @dok2001 - **mjackson** (2025-05-30): https://x.com/mjackson/status/1928566569774567480 - Ah, the venerable <iframe>! Relegated to the sidelines. Forgotten. Mocked! An ancient relic, doomed to an inconspicuous existence in the most obscure annals of history. … or is it? https://t.co/HOXAlpaqkz - **soleio** (2025-05-30): https://x.com/soleio/status/1928526076944248866 - A new show for you to binge this weekend. We call it: FIRST OF KIND Six episodes. Six intimate conversations with design pioneers. Season 1 is available: firstofkind • com https://t.co/IiGbmBOLCu https://t.co/PVRmT0UMFy - **vlucas** (2025-05-30): https://x.com/vlucas/status/1928512722078924813 - It’s funny reading the reactions to this. So many JS devs have no idea how simple working on websites used to be and how much we’ve lost. They literally can’t even imagine how this could be better than what we have now. https://t.co/qxvrBtZl7E - **scorecardai** (2025-05-30): https://x.com/scorecardai/status/1928482068717060567 - Announcing the first remote Model Context Protocol (MCP) server for evaluation! Built with @stainlessapi and @clerk to bring Scorecard evaluations directly into your AI workflows. Learn more: https://t.co/eAuWveckRh - **CloudflareDev** (2025-05-30): https://x.com/CloudflareDev/status/1928464666650656776 - Register (it’s free!) and drop by the @Cloudflare offices in SF next Monday to talk about modern data engineering! 👇 https://t.co/HD98d316PE - **benallfree** (2025-05-30): https://x.com/benallfree/status/1928434584582332760 - Day 3 of working on AdminDO, the admin for @Cloudflare durable objects - making the `admindo` CLI tool Last time, we were refactoring dofs to export hono middleware so dofs could have an HTTP API. Then, we updated dterm to use that API endpoint. Today we're continuing by wrapping up this refactor, releasing dofs and dterm official releases, and making the admindo CLI client with dterm support. - **mjackson** (2025-05-30): https://x.com/mjackson/status/1928420584179769592 - What if web dev tooling were optimized for browsers instead of build pipelines? What would that look like? - **JoviDeC** (2025-05-30): https://x.com/JoviDeC/status/1928357373837549907 - Let me also take this opportunity with Preact being a hot-topic now that I write about Preact topics on my blog! If you want to know more about the new diffing algorithm, you can look at the source code and read a few of my articles! - **mjackson** (2025-05-29): https://x.com/mjackson/status/1928215819374252450 - CoffeeScript walked so JSX could run - **MikeNomitch** (2025-05-29): https://x.com/MikeNomitch/status/1928157418292400542 - If you are interested in PMing on Containers or Workers, let me know! - Looking for people in Austin, London, Lisbon, SF, or NYC. Open to EMs or engineers that are thinking about making the jump to PM. Technical background very much appreciated! https://t.co/M47815nlj0 - **corywilkerson** (2025-05-29): https://x.com/corywilkerson/status/1928108222113796619 - Convinced that all problems can be solved with a @cloudflare worker. - **CloudflareDev** (2025-05-29): https://x.com/CloudflareDev/status/1928099205047873628 - Want to learn about how we built the new Workers Builds CI/CD system using Durable Objects? Read on 👇 https://t.co/cViHxFvNR0 - **ritakozlov** (2025-05-29): https://x.com/ritakozlov/status/1928096778802356229 - https://t.co/hfxbmKDfcm - **ritakozlov** (2025-05-29): https://x.com/ritakozlov/status/1928096776537456734 - we built our builds system entirely on @cloudflaredev: workers, durable objects, hyperdrive (to connect to postgres), containers & more customers often ask us about best practices, so we wanted to share how we solve real problems on the platform great post from @jachands 👇 - **RozenMD** (2025-05-29): https://x.com/RozenMD/status/1928086075928445016 - @sergioxro @dok2001 @Cloudflare @ritakozlov @elithrar @williamallen @irvinebroque re: A better template page, maybe even with the option of instant deploy working on that atm - the templates already integrate with the deploy to cloudflare button - **sergioxro** (2025-05-29): https://x.com/sergioxro/status/1928080723489038741 - @dok2001 @Cloudflare @ritakozlov @elithrar @williamallen @irvinebroque For the next PMs: A better template page, maybe even with the option of instant deploy A way to don’t kill the workers if client disconnects pls Better explanation on how to use wrangler types and a suggested flow Copy past env vars “Migrate from” tutorials - **dok2001** (2025-05-29): https://x.com/dok2001/status/1928078769987653919 - If you want to be a PM on @Cloudflare Workers platform please DM @ritakozlov @elithrar @williamallen @irvinebroque or anyone else on the team. We have a lot to ship 🚢 this year and need help making the platform delight developers everywhere! - **pcattori** (2025-05-29): https://x.com/pcattori/status/1928067230413246798 - Longer answer in replies, but tl;dr: Remix 3 won't assume Vite is there, but we want to let you layer it on top for DX and prod optimizations. The Remix 3 way: Bundlers should optimize features, not enable them. https://t.co/OHpSfrZkNJ - **KenjiCapital** (2025-05-28): https://x.com/KenjiCapital/status/1927843150380552415 - Two new towers proposed in San Francisco.... They'd be 58-stories and 51-stories. The lot was demo'd in 2020 and has sat vacant since. This project would bring: • 1,500 apartment units • 468 parking spaces • 2,300 sq ft of retail 📍San Francisco (SOMA) https://t.co/XNgNcmyff9 - **ritakozlov** (2025-05-28): https://x.com/ritakozlov/status/1927817179853930998 - dev.vars haters, this one's for you 👇 https://t.co/Bwot6PU3UF - **korinne_dev** (2025-05-28): https://x.com/korinne_dev/status/1927815811139285494 - We want your feedback! Here's a proposal to support loading environment variables via .env files for Cloudflare Workers: https://t.co/s93ZblYlra This has been highly requested as an alternative to using .dev.vars, and we're excited to get your thoughts. - **ShopifyEng** (2025-05-28): https://x.com/ShopifyEng/status/1927777585645191508 - Import maps and JavaScript modules are important web platform primitives for performance benefits 🐎 but not reliable at scale. So we fixed it! 🎉 Learn how in this post by @yoavweiss https://t.co/ouz812EoTT - **gakonst** (2025-05-28): https://x.com/gakonst/status/1927764579028783156 - Introducing the Porto Developer Preview! What if passwords did not exist? What if the web was built natively for payments? Porto is the open-source Typescript library we've built @ithacaxyz that gives accounts the superpowers to answer these questions. Learn more below 🧵 https://t.co/vXcM2uVXzq - **remix_run** (2025-05-28): https://x.com/remix_run/status/1927729823805255925 - Wake up, Remix! https://t.co/0ZO76NUqGC - **eastdakota** (2025-05-28): https://x.com/eastdakota/status/1927703382136852848 - Something I just told a journalist which may not make the story: I think one of the best things Portugal should do is *raise* taxes on unoccupied/part-time occupied real estate. It’s crazy that 50% of Lisbon real estate sits vacant. Want to make housing cheaper? Make it expensive to sit on real estate and not make it productive. I own a home in Lisbon I only live in part time. I believe I should pay more real estate taxes as a result. And the totally vacant properties that could be used for more supply of homes and apartments should too. If it’s free to hold onto real estate without making it productive that’s what people will do. And it’s why Portugal had fewer new housing starts in the last 5 years than any other developed country. - **jldec** (2025-05-28): https://x.com/jldec/status/1927577998900748617 - Realtime streaming RSC with ⁦@RedwoodJS⁩ on ⁦@CloudflareDev⁩ https://t.co/NOjjnZLpNm - **benallfree** (2025-05-28): https://x.com/benallfree/status/1927549103791108161 - D4 EOD: making @pocketbase work with @Cloudflare containers. We got PocketBase stdout/stderr streaming via websocketd back to the Durable Object. This is huge news because: a) it means we can talk via websocket to the container b) it means we can see process stdout/stderr output from running processes c) it means we can tail pocketbase console output via CLI and right to the pockethost .io dashboard i'm not sure what cf had in mind for accomplishing this level of integration, but it's super important for pockethost. if they didn't have anything planned...this will work :) - **yusukebe** (2025-05-28): https://x.com/yusukebe/status/1927537475641294878 - Cloudflare Agents SDK x Hono x Drizzle is neat. You can use D1 inside the MCP tools. https://t.co/j4fWiPMMhv - **garrytan** (2025-05-27): https://x.com/garrytan/status/1927497884754407933 - I just got my season tix for the Warriors next year, and I wasn't looking forward to having to deal with all the ticket distribution and/or sales for games I can't make. This is a cool use of AI — making a crufty workflow easy that can turn into long term retention. https://t.co/jajMuSG1BG - **GergelyOrosz** (2025-05-27): https://x.com/GergelyOrosz/status/1927285131380732302 - Thomas's reply: those companies have it backwards. If you're not hiring young people who are open-minded about new ideas (including GenAI) then you risk getting stuck where you are, innovation drops. Plus, young folks (interns/new grads) often improve team dynamics greatly. - **donvito** (2025-05-27): https://x.com/donvito/status/1927262683100782930 - wow, didn't know this cloudflare has an Agents SDK too has anyone tried this? https://t.co/pQvIyw5Sw2 - **acoyfellow** (2025-05-26): https://x.com/acoyfellow/status/1927009066133467397 - Say hello to userdo A tiny RPC layer for Durable Objects. No routers. No boilerplate. Just drop in a class and call methods from anywhere. Built for Cloudflare. Feels like magic. - Install it - Wrap your class - Call remote methods like local ones @CloudflareDev #Cloudflare https://t.co/1wX5HoLo7k - **eastdakota** (2025-05-25): https://x.com/eastdakota/status/1926752072315060469 - @joehas I have no idea. I’d guess they’re humbled as the tax advantages have gone away and the food and weather still sucks. Maybe we’ll reevaluate. - **amasad** (2025-05-25): https://x.com/amasad/status/1926709193404198968 - @arblauvelt Product engineering and design - **Cloudflare** (2025-05-25): https://x.com/Cloudflare/status/1926688678471999701 - Why is everyone talking about MCP and AI agents? In this #ThisWeekinNET episode, we cover with @nevikashah and @dinasaur_404: ✅ What MCP is ✅ Demos on Cloudflare ✅ Why companies should care ✅ What’s next 🎥 Watch now: https://t.co/2GVqUYnm74 https://t.co/4zk0yHjeit - **amasad** (2025-05-25): https://x.com/amasad/status/1926646068726440357 - After 3 years of remote, 2 years of hybrid, and a year of mostly in-office—the team that can be most distributed in a modern startup is infra engineering. Perhaps because the requirements are knowable, systems need to change more deliberately, and you benefit from writing everything down. - **thdxr** (2025-05-25): https://x.com/thdxr/status/1926630878752526482 - every few months i consider moving terminal to shopify on the backend it's normally how i'd do things - it's dumb to recreate the infinite pool of e-commerce functionality but whenever i go to look at it, it's clear that we're just a little too weird to properly work in their system without a ton of hacks/plugins i want something lightweight and modular - so i think hm is there an opportunity to build this? the answer is no - they built in order of demand and if my use case does not fit their product, it's probably pretty uncommon it's old software and it doesn't feel great to use - but even then there's no opportunity are you really going to spend the next 10 years trying to keep yourself afloat building the massive set of features to replace shopify? you think maybe there's some small slice of features that'll work for some customers but it's always still a big list so we're kind of forever stuck with these things until there's some major paradigm shift that opens up an angle of attack - **liadyosef** (2025-05-25): https://x.com/liadyosef/status/1926626307229520019 - Wow, this is awesome!! @OpenAI chose GitMCP as the official example in their MCP integration docs! 🤯 Check it out: https://t.co/kvxXTXDAqK (+@idosal1) https://t.co/STXpHJi0lh - **davidhoang** (2025-05-24): https://x.com/davidhoang/status/1926416138667282893 - We found this guy stuck in the car engine yesterday. He was so scared. Since then he went to the vet to get checked up and now hanging at our house for now. Right now we’re using Wilson’s old blankets and toys to make him feel safe. https://t.co/MaXnDuVV11 - **dinasaur_404** (2025-05-24): https://x.com/dinasaur_404/status/1926128220828565704 - best part of mcp dev summit was getting to meet everyone that’s working on making mcp better :) https://t.co/t9SZFgfgQ2 - **aarmlovi** (2025-05-24): https://x.com/aarmlovi/status/1926118257490112690 - Zohran *tonally* moderated, but under the familiar *thematic* jargon he's hard-line opposed -He openly lied about the RGB "hardship increase", knowing Joe & Tracy wouldn't know it -He only supports streamlining housing that meet "affordability, labor, & energy standards" -Single stair is already legal in NYC! And he does not support expanding it -He opposed the MCI increase, one of the only fixes to distressed rent stabilized housing that actually rewards investment instead of speculation - **techgirl1908** (2025-05-23): https://x.com/techgirl1908/status/1926058350149566657 - .@kentcdodds was amazing at #MCPDevSummit today! 🔥 https://t.co/aGsm4Si7qm - **kentcdodds** (2025-05-23): https://x.com/kentcdodds/status/1926052041870168369 - Excited to hear about building excellent MCP servers from @dinasaur_404 at #MCPDevSummit https://t.co/lEV3E7XxFR - **zebassembly** (2025-05-23): https://x.com/zebassembly/status/1926049171741569368 - Today was a VERY good day - **dsp_** (2025-05-23): https://x.com/dsp_/status/1926029319047958663 - A special moment. A bunch of people from the MCP Steering Committee getting together for the first time. https://t.co/DrLbpmMNN3 - **bboyle18** (2025-05-23): https://x.com/bboyle18/status/1926012785802047537 - Big week @intercom. We just announced Fin Insights: a groundbreaking product that gives you complete visibility into every customer conversation, with AI-powered tools and suggestions that help you monitor, analyze, and instantly optimize your customer service quality. 👇 https://t.co/5rRRpiyLdC - **paulpopus** (2025-05-23): https://x.com/paulpopus/status/1925971496863776986 - We just deployed @payloadcms to @Cloudflare with D1 database and R2 for cache and storage. Performance is really good, we're gonna release a template in beta for this soon There's still a lot for us to do but this is extremely promising, thank you to the CF team working through issues! Thanks to @r1tsuu__ for his work on the db adapter - **kentcdodds** (2025-05-23): https://x.com/kentcdodds/status/1925959808831647925 - Up now at #MCPDevSummit is @llmindsetuk talking about mcp resources. A underutilized part of mcp! https://t.co/xBlGx3EC2g - **eastdakota** (2025-05-23): https://x.com/eastdakota/status/1925948418712609263 - This is what downtown SF needs to focus on more than anything. Make it the easiest place to open a restaurant, bar, coffee shop, specialty bodega, or any other sort of shop! https://t.co/dsYxvho9CQ - **liadyosef** (2025-05-23): https://x.com/liadyosef/status/1925945287417143403 - Excited for the @mcpsummit! Met the amazing @kentcdodds, @dinasaur_404 and @tadasayy this morning and already chatted about the future of MCP, looking forward to the great talks today! (and to meeting @techgirl1908, @dsp_ and @evalstate in person) https://t.co/pPSts5Scx7 - **jeff_weinstein** (2025-05-23): https://x.com/jeff_weinstein/status/1925944830493852095 - Maybe a long shot, but is anyone building an ai-first “word processor”? I would love to play with it, even if very early and help you. - **ycombinator** (2025-05-23): https://x.com/ycombinator/status/1925914957968773439 - Many AI apps today feel like the "horseless carriages" of the late 19th century, which "swapped a horse for an engine without redesigning the vehicle to handle higher speeds." They pack powerful tech into outdated interfaces. YC's Pete Koomen (@koomen) thinks we can do better. He joined @t_blom and @dflieb to lay out a new vision for how AI should actually work: not as a chatbot bolted onto legacy software, but as a customizable tool that helps people offload the work they don't want to do. From editable system prompts to agents that act more like collaborators, they break down what it means to build AI-native software—and why the future belongs to products that let users teach machines how they think. 0:00 – Intro 0:52 – Why AI apps are broken 2:39 – How Gmail’s AI features fall short 4:00 – A better way to build AI apps 5:27 – The hidden system prompt 7:57 – What if you could access the system prompt? 9:40 – The developer-user divide in software 10:48 – The "horseless carriage" metaphor 13:35 – Email reading agent demo 14:34 – Everyone can be a prompt engineer 16:23 – Why coding agents feel magical 21:42 – Training AI like a human assistant 28:45 – The problem with chatbot interfaces 29:10 – Advice for founders - **emot** (2025-05-23): https://x.com/emot/status/1925877846632731065 - Why is everyone talking about MCP and AI agents? In this #ThisWeekinNET episode, we cover with @nevikashah and @dinasaur_404: ✅ What MCP is ✅ Demos on Cloudflare ✅ Why companies should care ✅ What’s next 🎥 Watch now: https://t.co/H56i2Qj0GP @CloudflareDev @Cloudflare https://t.co/Epo5R920Rn - **AlecStapp** (2025-05-23): https://x.com/AlecStapp/status/1925861607906668921 - Not sure how many people realize that battery storage already supplies 30% of California’s electricity demand at peak. Batteries aren’t the future — they’re here now. https://t.co/p4DwRWg9vm - **kentcdodds** (2025-05-23): https://x.com/kentcdodds/status/1925762167220957400 - I'm excited I got to meet and chat with my new friends @dinasaur_404 and @dsp_ at the @mcpsummit speaker dinner! https://t.co/MHktsriItD - **thdxr** (2025-05-23): https://x.com/thdxr/status/1925746936587399258 - @AdamRackis ok so given that is your first thought, what is your second thought? - **samgoodwin89** (2025-05-23): https://x.com/samgoodwin89/status/1925743939606175974 - Alchemy hit 400 stars today. Growing steadily, but hard earned! Please star the repo if you haven't already 🤗 (link in replies). https://t.co/LBMXG0gvce - **thdxr** (2025-05-23): https://x.com/thdxr/status/1925738988138049738 - take at the look at the replies everyone concluding the people who spent at minimum a year thinking about this decision are complete idiots watch what happens next https://t.co/moZvlkUw0z - **mjackson** (2025-05-23): https://x.com/mjackson/status/1925720267348660696 - Haters gonna hate. Just keep shipping. - **kentcdodds** (2025-05-23): https://x.com/kentcdodds/status/1925705340030431288 - I'm here to speak and meet people at @mcpsummit. Got a ride into San Francisco by this dude called BART. Gonna be a good couple days! https://t.co/YTggKlWv2G - **boristane** (2025-05-22): https://x.com/boristane/status/1925653150251294852 - we're fast approaching a world where you can vibe code a whole suite of internal tools, tailor them to your specific needs instead of buying big SaaS-industrial complex - **samuelgoddard** (2025-05-22): https://x.com/samuelgoddard/status/1925627103141097816 - Absolutely massive shift put in by the entire team @intercom over the past months creating a brand new site from the ground up for https://t.co/6bFhJ9lAPd - huge tech stack upgrades including @nextjs app router and much more, learnt a tonne and super proud of what we shipped 🚀 - **davidhoang** (2025-05-22): https://x.com/davidhoang/status/1925553695464513980 - Millennials were built for Agent Orchestration. https://t.co/sZFNcd2VrF - **brandinpodziem2** (2025-05-22): https://x.com/brandinpodziem2/status/1925410388398780665 - Chase center was jumpin tonight oh my god!!!!!!!! - **ChaseCenter** (2025-05-22): https://x.com/ChaseCenter/status/1925407592433815986 - FIRST-EVER @valkyries WINNN!!! https://t.co/38MFRoQZsV - **CaitlinClark22** (2025-05-22): https://x.com/CaitlinClark22/status/1925384528874332581 - PACERS ARE THE GREATEST COMEBACK TEAM IVE EVER SEEN - **KevinOConnor** (2025-05-22): https://x.com/KevinOConnor/status/1925369501542916378 - These are the whistles they gotta stop https://t.co/CLuxT6BQCk - **michellechen** (2025-05-22): https://x.com/michellechen/status/1925362775112421455 - knicks in 5?? https://t.co/N4U9J12o4w - **thdxr** (2025-05-22): https://x.com/thdxr/status/1925349029069611091 - @naxodev we're actually adding DO's natively to sst.cloudflare this week we're finally using them ourselves - **stevenheidel** (2025-05-21): https://x.com/stevenheidel/status/1925340416049901847 - we built the MCP + Responses API integration in record time, largely thanks to Codex. i personally kicked off over 50 tasks and merged thousands of lines of Codex-written code - **sergiodxa** (2025-05-21): https://x.com/sergiodxa/status/1925339864989421736 - I think I finally get @CloudflareDev Durable Objects. If Workers are serverless functions, then DOs are serverless objects. Workers run and disappear. DOs stay in memory, so they can hold state instead of pushing everything to a DB. - **sergiodxa** (2025-05-21): https://x.com/sergiodxa/status/1925332761398710302 - Tried @cloudflaredev Durable Objects for the first time and used them to power SSE from a @remix_run React Router route Now I get real-time updates when an uptime check finishes and can trigger a revalidation Here’s it in action 👇 https://t.co/GDQLklJCHU - **kimmaicutler** (2025-05-21): https://x.com/kimmaicutler/status/1925290348571701313 - We really wasted ZIRP when we could’ve been building all kinds of infrastructure, energy and housing. https://t.co/mdBWE9PRoV - **FredKSchott** (2025-05-21): https://x.com/FredKSchott/status/1925272843627160028 - Looks like Next.js App Router doesn't meaningfully improve Core Web Vitals perf in real-world usage. https://t.co/MJYcRwbA6l - **stevenheidel** (2025-05-21): https://x.com/stevenheidel/status/1925218271349104851 - you're now just one API call away from connecting with hundreds, and soon thousands, of tools across the web https://t.co/JxOdEQxdjd - **craigsdennis** (2025-05-21): https://x.com/craigsdennis/status/1925213646763991174 - 💪 WHOA! Y'all gotta check this out! I built a quick demo to make use of the new MCP tool in the @OpenAIDevs Responses API https://t.co/2h3OBZf9j3 https://t.co/ZNx7jTHNfi - **kevinweil** (2025-05-21): https://x.com/kevinweil/status/1925211565780873498 - The OpenAI Responses API now supports MCP! With just a few lines of code you can connect OpenAI models to any remote MCP server. More to come. https://t.co/fv5hnh87yZ - **realEstateTrent** (2025-05-21): https://x.com/realEstateTrent/status/1925188042761052369 - I’m in total disbelief. I left this city a few years ago after spending my entire adult life here. I’d given up on the idea that what I’m seeing right now would be possible in my lifetime. I’m shocked to type this, but has San Francisco become the cleanest big city in America? https://t.co/bagFUxRWPi - **charl_dot_dev** (2025-05-21): https://x.com/charl_dot_dev/status/1925183993684673002 - MCP Hub - Cloudflare remote MCP server that lists other MCP servers, surfaces their tools, props etc and can execute their tools. e.g. const DOWNSTREAM_MCP_URLS = [ { id: "calc", url: "http://localhost:63986/sse" }, ]; https://t.co/1DaeO8vWRo - **expo** (2025-05-21): https://x.com/expo/status/1925175257738187110 - 🪲 Finding and fixing bugs in production just got easier for Expo developers. You can now see @getsentry issues and session replays directly in your EAS deployments. Here’s how it works—and why it matters: 🧵 https://t.co/9pPlnWAePA - **elithrar** (2025-05-20): https://x.com/elithrar/status/1924947432015609932 - Imagine it, then build it on Cloudflare Workers 🧠 https://t.co/Bc2GF4r06U - **gingerhendrixai** (2025-05-20): https://x.com/gingerhendrixai/status/1924936820447392062 - Git on a Durable Object Clone or init a git repo into a durable object. It has a working copy so you can edit, commit, branch, push etc. This is a side-project unblocker for me. You could use it to build your own deepwiki, chat with repo, coding agent, collaborative editor? https://t.co/ttzKuWuQaa - **jonas** (2025-05-20): https://x.com/jonas/status/1924875751624524088 - @BraydenWilmoth Yesssss! Please make DurableObjectState work a bit like Agent state An object. No more individual KV putting! And also make it so I can use handler(MyActor) easily within another fetch handler - **BraydenWilmoth** (2025-05-20): https://x.com/BraydenWilmoth/status/1924872988370821140 - In a world..... where this is the minimum it took to define a Durable Object.... Coming soon to an IDE near you. No worker. Just a class. Export through a handler to route traffic to it. https://t.co/6uBzjgVJjK - **dinasaur_404** (2025-05-20): https://x.com/dinasaur_404/status/1924818939156299867 - MCP only business built using @Cloudflare for the new streamable http transport (peep the /mcp), auth, and mcp-remote + @stripe for payments! 🙌 https://t.co/hae5ICIKka - **iannuttall** (2025-05-20): https://x.com/iannuttall/status/1924801656765526066 - my mcp server boilerplate is now free + open source (just refunded all orders) - auth users with google/github - host on cloudflare for free - one-time, subscription & metered billing with stripe pull requests welcome - lets build some profitable mcp servers - link below ↓ https://t.co/a4ygrN9JZy - **sqs** (2025-05-19): https://x.com/sqs/status/1924604926883463256 - MCP done right — What are the best-designed MCP servers, the ones with good tools (at the appropriate level of abstraction) and descriptions, good response text, good impls of prompts and resources, etc.? Please help me add to this list: - Playwright - PostgreSQL - Sentry - Linear - **zebassembly** (2025-05-19): https://x.com/zebassembly/status/1924587098436403641 - Something that isn't brought up enough (but still something to think about for framework authors) is the fact you can load/render different parts of a request in different data-centers on Cloudflare without any additional cost. With Cloudflare and frameworks with smart bundler tricks, the wisdom of "the edge is slow because the database is far away" can go away. - **juliannaelamb** (2025-05-19): https://x.com/juliannaelamb/status/1924539508789543395 - We're hosting an MCP meetup with Cloudflare on Thursday! Come learn how to build and deploy a secure, remote MCP server with in depth tutorials from @lizziepika and one of our engineers, Max https://t.co/BNhejLPJ7g - **MathurAditya7** (2025-05-19): https://x.com/MathurAditya7/status/1924521874333303170 - Now you can deploy the Muppet MCP Inspector on @Cloudflare https://t.co/kjcz3jfeGm - **danshipper** (2025-05-19): https://x.com/danshipper/status/1924516634582253572 - Microsoft CTO Kevin Scott compares MCP to HTTP - **cjpatton_** (2025-05-19): https://x.com/cjpatton_/status/1924479593047863585 - I work with some pretty spectacular people. - **ohnohedidnt24** (2025-05-18): https://x.com/ohnohedidnt24/status/1924244450462597449 - Aaron Gordon: "I would really, really appreciate it if there were a couple of days in between games in the playoffs instead of every other day. The product of the game would be a lot better. You'll see a higher level of basketball. Probably less blowouts" https://t.co/oLxKiEG95N - **KnicksMuse** (2025-05-17): https://x.com/KnicksMuse/status/1923616569524990131 - CHALAMET https://t.co/1MjwwtCOVP - **Rachel__Nichols** (2025-05-17): https://x.com/Rachel__Nichols/status/1923570866811699330 - Knicks-Pacers with Reggie Miller on the call - the final NBA series TNT will broadcast as their current TV deal expires. What a fitting way to go out. - **idosal1** (2025-05-16): https://x.com/idosal1/status/1923477857881190718 - Introducing MCP UI, the SDK that wires UI components directly into the MCP protocol🤯 Watch the video to see the new phase in AI interaction! Why browse when you can have universal apps that deliver the right UI to consume any data and perform any action? Replace cumbersome walls of text with slick interactive components, bringing rich web experiences to MCP-enabled agents such as Claude and Cursor. Let’s build Jarvis so @kentcdodds doesn’t have to! 🧵>> - **majek04** (2025-05-16): https://x.com/majek04/status/1923387389877096703 - Today, after almost 12 years, is my last day at Cloudflare. I've seen the company grow from 60 to 5,000 people. I helped bootstrap things like DDoS, Spectrum, Cloudchamber, and Egress, and wrote hundreds of internal and dozens of public blog posts. It has been an amazing journey. https://t.co/XkM3XA8LZq - **samlambert** (2025-05-16): https://x.com/samlambert/status/1923353304916000931 - this is like 5 postgres hosting companies worth of traffic coming from just one of our customers. https://t.co/m5q7fuNifM - **ryan_sherlock** (2025-05-16): https://x.com/ryan_sherlock/status/1923335948147769478 - A week later, and we've another million RPS moved across to @PlanetScale. This represents the majority of the database load in our US region now migrated across. https://t.co/ucoNfUIaWl - **SFBART** (2025-05-15): https://x.com/SFBART/status/1923116904589250708 - Great work by BART Officer Saechao and BART Police K9 Baron assisting Oakland Police by sniffing out this handgun. On May 6, 2025, at around 5:30pm, BART Police K9 Baron sprang into action following a shooting near 14th Street and Broadway in Downtown Oakland. Thread ⬇️ https://t.co/rnV3dQAa2I - **theo** (2025-05-15): https://x.com/theo/status/1923106496922955840 - Underrated part of having a lot of devs using your product: the bug reports are so detailed and actionable - **alanaagoyal** (2025-05-15): https://x.com/alanaagoyal/status/1923105219321188640 - it's crazy to me that slack isn't an mcp client yet - **ritakozlov** (2025-05-15): https://x.com/ritakozlov/status/1923095320168202355 - about to have every exec at @cloudflare vibe code a workers app ✨ - **eastdakota** (2025-05-15): https://x.com/eastdakota/status/1923054116713922662 - This is incredibly important for the future of an agentic web. If bots are going to do more for us, we need to reliably identify them. Here’s the open standard we’re working with the IETF to implement. https://t.co/kkoPdOuIkr - **elithrar** (2025-05-15): https://x.com/elithrar/status/1923053236786692522 - One of the best things about working @Cloudflare is the Internet-wide standards work we do at scale. really proud of the work Mari, Thibault & the team did to make cryptographically verifying bots and AI agents a reality: https://t.co/r6g2qHwmIL - **CloudflareDev** (2025-05-15): https://x.com/CloudflareDev/status/1923048959276089691 - New Workers Analytics Engine UI just dropped 🙌 https://t.co/y8WkGOWp8s - **TweetAnnaMarie** (2025-05-15): https://x.com/TweetAnnaMarie/status/1923039001218273775 - So great to vibe with you all, @TobinSouth, @biilmann, @jaredhanson, and @irvinebroque 🙌 - **_ashleypeacock** (2025-05-15): https://x.com/_ashleypeacock/status/1923034295947391016 - Native Python Workers on Cloudflare now support Durable Objects 🚀 Continuing a running theme of @CloudflareDev Python Workers getting more features, fresh off the back of cron triggers being added recently, you can now build Durable Objects with Python natively on Cloudflare! https://t.co/kAD7KFlF4e - **thdxr** (2025-05-15): https://x.com/thdxr/status/1923007567665438868 - old but true wisdom is the only thing that matters is building your product and talking to users this has never been harder because the number of productive seeming distractions available to you is at an all time high - **yusukebe** (2025-05-15): https://x.com/yusukebe/status/1922967470090092706 - みんなでCloudflare Workers上にリモートMCPを建てるワークショップってどうですか？ - **grinich** (2025-05-15): https://x.com/grinich/status/1922909665689501999 - wow what an INCREDIBLE night! thank you to all the amazing presenters, panelists, and attendees! feeling especially grateful to the awesome team at WorkOS who absolutely hit a home run 🖤 more photos and videos soon! https://t.co/ccx9rNw6cZ - **janwilmake** (2025-05-15): https://x.com/janwilmake/status/1922903327597019310 - Update: stripeflare v0.0.4 - uses crypto to encrypt and decrypt a client_reference_id directly from the access_token - result: more efficient, does not create user until payment How? https://t.co/q0XwwsBwDe https://t.co/mNR9QJvDzu - **reed** (2025-05-15): https://x.com/reed/status/1922892574068519065 - cute blue @waymo https://t.co/uedSjsVcNK - **957thegame** (2025-05-15): https://x.com/957thegame/status/1922889850140741769 - “They pointed that out last year. And the year before, too. One thing I can assure them is every year we will get older… I don’t worry about that one bit. I feel like I’m still improving. I feel like Steph is still improving. I feel like Jimmy’s still improving.” Draymond on people talking about the age of the Warriors Big 3 - **holman** (2025-05-15): https://x.com/holman/status/1922835296200061243 - honestly hecka cool to watch @cjbell_ live demo @knocklabs while using @raycastapp at the standing-room-only @WorkOS MCP Night event tonight, yes I invested in all three of those startups but that’s besides the point, they’re all awesome https://t.co/ma5RZVIbov - **chantastic** (2025-05-15): https://x.com/chantastic/status/1922832857682018790 - KylieOS✖️ @Cloudflare @kyliestew #MCPNight https://t.co/wgI2YaLzKr - **chantastic** (2025-05-15): https://x.com/chantastic/status/1922827687405494707 - 😄 @grinich “who knows all the details of the OAuth spec?” …only @aaronp raises their hand #MCPNight - **ajoshhoward** (2025-05-15): https://x.com/ajoshhoward/status/1922820401056354537 - Thanks for having me on @aarondfrancis! I’m a big fan of the podcast… and they say “Don’t meet your heroes”…. It was fun 🙂 https://t.co/8yuvDJifQa - **grinich** (2025-05-14): https://x.com/grinich/status/1922782958487818524 - see you in a couple hours! https://t.co/Cph2AEWfOG - **benblumenrose** (2025-05-14): https://x.com/benblumenrose/status/1922773990411575795 - Some wonderful insights at @spc from @colindunn and @soleio around running a company as a designer founder - “Make sure you love the problem space you’re in - it’s easy to fall in love with the idea of starting a company but that’s not enough.” https://t.co/ag2npqHnA0 - **FredKSchott** (2025-05-14): https://x.com/FredKSchott/status/1922730034407801164 - What if your inbox could manage itself? I've been trying to crack this problem for a while alongside @matthewcp + a few others, and we finally did it. Introducing Marlo: an intelligent email client that organizes itself for you. Waitlist now available! Link in replies. - **JustJake** (2025-05-14): https://x.com/JustJake/status/1922713542752338088 - A workflow I'm finding works well I ask Cursor to do something It finds it in the codebase It botches the implementation I, now having all the context, implement it myself It's kinda better at Search than it is at code completion - **lizziepika** (2025-05-14): https://x.com/lizziepika/status/1922709099646513372 - Peep the @CloudflareDev booth @ @langchain Interrupt--hack the safe like @FanaHOVA && you can win this pacman arcade lego! (and an Agents hat || shirt) https://t.co/s4rKQz42WW - **zeeg** (2025-05-14): https://x.com/zeeg/status/1922697398607831043 - Trace-connected Logs are now in Open Beta on Sentry! We're still iterating on the product experience, but we'd love for you to give it a try and tell us what you think. This also gives you a path to send less-actionable data (warnings, or even some errors) as logs vs issues 💪 https://t.co/1gGRcmysvS - **aarondfrancis** (2025-05-14): https://x.com/aarondfrancis/status/1922683060769685665 - I sat down with @ajoshhoward from @Cloudflare and asked him my most pressing question... What *is* a Durable Object? We spent the next hour talking Durable Objects, D1, Workers, and how the whole Cloudflare platform works. You're gonna love this one. https://t.co/uSomsj91es https://t.co/Fi2fbLMzHk - **HoodiiMamba** (2025-05-14): https://x.com/HoodiiMamba/status/1922450598173024350 - Oracle was a devils pit man https://t.co/u8gT74cfhg - **erikdunteman** (2025-05-14): https://x.com/erikdunteman/status/1922446181281550439 - Announcing: Muscle Mem 💪 Muscle Mem is a cache system for AI agents, allowing them to learn and efficiently replay complex behaviors. This allows expensive LLM calls to be entirely removed from the hot path, during repetitive tasks. https://t.co/xmMMRpqsY7 - **mjackson** (2025-05-13): https://x.com/mjackson/status/1922413984507957672 - Server-driven UI is awesome. Complex bundler magic is not. You don't need the latter to get the former. - **FredKSchott** (2025-05-13): https://x.com/FredKSchott/status/1922413757344416194 - For the past few months, I've been secretly working on a new email client. I think it's finally ready to share with the world... https://t.co/RDZo6EsvkH - **ycombinator** (2025-05-13): https://x.com/ycombinator/status/1922411531851202882 - Better Auth (@better_auth) is a comprehensive authentication framework for TypeScript that helps you roll your own auth from simple flows to enterprise-grade systems directly on your own database, embedded in your backend. https://t.co/5iCHUxZmZL Congrats on the launch, @bekacru + @KinfishT! - **kentcdodds** (2025-05-13): https://x.com/kentcdodds/status/1922321790136668410 - HTTP Server : Website :: MCP Server : ? - **ritakozlov** (2025-05-13): https://x.com/ritakozlov/status/1922310128323461552 - a misconception about @cloudflaredev is it runs your code everywhere not anywhere everywhere is good for latency-sensitive use cases (near the user) but most data-rich applications should be near data — anywhere region: earth just means the platform takes care of this for you - **burcs** (2025-05-13): https://x.com/burcs/status/1922303282891858366 - Started writing down how we think about DevEx at Cloudflare. Turns out it's just a continuation of the philosophy behind Outerbase. https://t.co/sIvY7B73gP - **NatalieMarleny** (2025-05-13): https://x.com/NatalieMarleny/status/1922274396128915517 - Gitlip Launch 🌷🚀 @GitlipDotCom is an all-in-one coding platform for shipping micro apps 👼💫 It features: 🌷Git repositories, 🤝 collaborative and ✨ AI assisted coding and ☁️ serverless deployments. No login required to try out! https://t.co/RaaTbyOIzm - **threepointone** (2025-05-13): https://x.com/threepointone/status/1922270689416851627 - agents@0.086 takes this one step further instead of agent. call('someFunction', [arg1, arg2]) you can now agent.rpc.someFunction(arg1, arg2) with typesafety! another banger from @ruifigueira https://t.co/iKMXwR1Roz https://t.co/E2eH0XM4EU - **daltonc** (2025-05-12): https://x.com/daltonc/status/1922066553761095988 - The S&P 500 now has 3 YC companies: DoorDash, Airbnb and Coinbase. - **TobinSouth** (2025-05-12): https://x.com/TobinSouth/status/1922024263902777832 - 800 people are gonna be at the Exploratorium on Wednesday to talk about making MCP work. Come along! https://t.co/0B4tl9bwRi - **korinne_dev** (2025-05-12): https://x.com/korinne_dev/status/1922009009403396406 - We also now have a guide for deploying a Redwood SDK (@RedwoodJS ) app on @CloudflareDev Workers! Thanks @appfactory for helping us get this up 🫶! https://t.co/tGUOD8ZHkj - **elithrar** (2025-05-12): https://x.com/elithrar/status/1921995960973987963 - this, 1000 times over. if you can ship faster, you can act on feedback faster. try new ideas faster. pivot faster. fix bugs faster. put too much process in the way and you lose the ability to execute on *all of it*. https://t.co/l3XSOopdVg - **mjackson** (2025-05-12): https://x.com/mjackson/status/1921937728003833960 - He replaced Next.js app router with a React Router SPA https://t.co/Lhuk7L8RJQ - **mcuban** (2025-05-12): https://x.com/mcuban/status/1921757618382446852 - The manufacturers aren't the big problem Bill. They could sell brand meds for far far less, and in some cases lower than other countries TODAY , IF you take the big PBMs out of the middle. And YOU are part of the problem Bill I'll bet you signed a pharmacy benefits deal with a big PBM? Tomorrow , call them and ask for a list of your claims, and the net drug price paid for each. Ask them for a net price list, after rebates that you can verify. Ask them if you can talk to say Lilly to discuss a GLP1 Wellness program. Ask them why they under reimburse pharmacies for less than their cost to help YOUR employees Let me know how long they laugh at you The issue is that CEOs have NO CLUE about their pharmacy or healthcare benefits. So they work with the PBMs that cause this mess, assuming their consultants are doing the right thing for them You enable the problem by working with those PBMs Bill . And of course its not just you. It's almost every big company CEO This EO can have a huge impact. But it has to be built around transparency and removing the middlemen - **lizziepika** (2025-05-12): https://x.com/lizziepika/status/1921737292365840822 - Doable in CA— just ran the #sbwinehalf Santa Barbara/solvang https://t.co/OC6v0cs8jT https://t.co/Cb7VQKP3mh - **martin_casado** (2025-05-11): https://x.com/martin_casado/status/1921679467547680799 - Programming with AI reminds me a lot of how I felt learning js: - There is a library/prompt for everything - It's never exactly what I want - I spend a lot of time trying to understand and debug someone/something else's code Yet the productivity gains are undeniable. - **CloudflareDev** (2025-05-11): https://x.com/CloudflareDev/status/1921574711936315584 - MCP gives @intercom's AI agents access to internal tools to look up rich customer data and history to resolve tickets faster. Be like @intercom, build your MCP server on Cloudflare https://t.co/M1XXnURIoe - **samgoodwin89** (2025-05-11): https://x.com/samgoodwin89/status/1921445170802733126 - Blogged about this in 2019, “imagining the future of cloud programming” where I envisioned a unification of runtime and infrastructure code. Alchemy finally enables this vision for typescript without overly complicated tooling. Blog in replies. https://t.co/kayH6K4hM2 - **samgoodwin89** (2025-05-11): https://x.com/samgoodwin89/status/1921440106713321574 - What if Alchemy Resources existed at runtime? I've been dreaming of this experience for over 5 years. https://t.co/3U7gTtlUXG - **ritakozlov** (2025-05-10): https://x.com/ritakozlov/status/1921307679466610848 - menu for dinner with friends tonight: - lemon sole in parchment w sautéed ramps - white asparagus with butter & lemon (hyped to have found white asparagus which i can never find in the us!!) - herby brown butter potato salad - rhubarb tart i love spring 🌸 https://t.co/jbcakAqFZI - **CloudflareDev** (2025-05-10): https://x.com/CloudflareDev/status/1921297491812651452 - To let you in on an even bigger secret: the Workers platform itself uses Analytics Engine for product & usage metrics! https://t.co/XqK7NhuvQW - **akothari** (2025-05-10): https://x.com/akothari/status/1921236550836166811 - SF’s downtown has taken a hit -- reports suggest vacancy is as high as 35%. But the story isn’t over. This week, @NotionHQ moved into a beautiful building that sat empty for years. Brick by brick, we can bring this city back. Let’s build the future here. Long San Francisco. https://t.co/QCqaJEl715 - **dsp_** (2025-05-10): https://x.com/dsp_/status/1921166770653528277 - @AstasiaMyers Please actually read the sources before posting something that is clearly not true. You can easily add auditing, and authorization is in the spec and people are actively using it day to day. - **brian_lovin** (2025-05-10): https://x.com/brian_lovin/status/1921045914518511795 - Had such an awesome and energizing Friday + week overall. I haven't had this much fun designing + making software in a long time. - **CloudflareDev** (2025-05-10): https://x.com/CloudflareDev/status/1921035670203588740 - Deploy your @tan_stack straight to Workers 👇👇👇 https://t.co/y3kx0oPmT1 - **korinne_dev** (2025-05-09): https://x.com/korinne_dev/status/1920991689981694453 - We've got a new guide for building applications on Cloudflare Workers with @tan_stack : https://t.co/uTt79tRCfz We're also working on getting a formal C3 template to scaffold much of this for you. - **nickbalestra** (2025-05-09): https://x.com/nickbalestra/status/1920983135346450790 - Little hat trick for YC folks: checkout https://t.co/K5Cc4Dwr1l to iterate on your @CloudflareDev stack even faster! https://t.co/dnvWTb2Vyt - **BraydenWilmoth** (2025-05-09): https://x.com/BraydenWilmoth/status/1920937722753011886 - Dark DARK mode is now live in Cloudflare. https://t.co/O2PNOvqxQO - **mwylde** (2025-05-09): https://x.com/mwylde/status/1920936020083671103 - @irvinebroque Five of us joined in April alone! - **ritakozlov** (2025-05-09): https://x.com/ritakozlov/status/1920934610235511186 - 🆕 new `enable_weak_ref` compatibility flag in @cloudflaredev workers runtime! this allows you to manage memory more efficiently when using webassembly (e.g. with python or rust) 🔗 changelog https://t.co/LpoE9BxMcW - **ycombinator** (2025-05-09): https://x.com/ycombinator/status/1920901582499045881 - Zero (@zerodotemail) is an AI-native email client that manages your inbox, so you don't have to. It's also fully open source, giving you transparency, control, and the freedom to make it your own. https://t.co/OVacEsJLr3 Congrats on the launch, @nizzyabi and @thecto! https://t.co/DVACLLLLKP - **thdxr** (2025-05-09): https://x.com/thdxr/status/1920870948611256748 - wow someone made a very thorough guide to using openauth with react router https://t.co/M11gpVQhjd - **ritakozlov** (2025-05-09): https://x.com/ritakozlov/status/1920867921393291330 - "what is @cloudflare's best kept secret? workers analytics engine" sorry @burcs for giving away your secrets 🤫 - **ryan_sherlock** (2025-05-09): https://x.com/ryan_sherlock/status/1920863089051779184 - In Dublin on May 22nd? wanna drop in to the Intercom office? Ruby Ireland are running a meetup in our office with (at least) one talk being about Rails at Intercom: https://t.co/EoIqNXl1ZZ https://t.co/wFJfQrczPZ - **ciaran_lee** (2025-05-09): https://x.com/ciaran_lee/status/1920860511173816765 - 59 production deploys of our rails monolith already today. Median e2e deploy time is ~11 mins. Fridays are shipping days here. - **ritakozlov** (2025-05-09): https://x.com/ritakozlov/status/1920856586764828831 - i was complaining recently that marketing does not come naturally to me, and a friend told me: "what's so hard? you just go: you need stuff we have stuff our stuff is the best stuff" my struggles have been solved ever since https://t.co/7n4kykrxh5 - **samseely** (2025-05-09): https://x.com/samseely/status/1920828076813861074 - Headed to SF with @cjbell_ next week. We'll be at @WorkOS MCP night Wednesday and in-out of @twilio Signal throughout the week. If you're in town would love to meet up and ☕ - **CloudflareDev** (2025-05-09): https://x.com/CloudflareDev/status/1920825496129921462 - .@asana's MCP server enables AI to become a true teammate in work management 👏 Be like @asana, build your MCP server on Cloudflare https://t.co/sYIGBwtmOz - **VinceGoodwill** (2025-05-09): https://x.com/VinceGoodwill/status/1920688368368726483 - Draymond Green with a short statement in the locker room following tonight’s loss, it starts with: “I am tired of the agenda to make me look like an angry black man…” https://t.co/k27Om6aOC3 - **craigsdennis** (2025-05-09): https://x.com/craigsdennis/status/1920643394872762487 - New video showing off the @CloudflareDev Agent Framework just landed showing off my favorite features: ✅ Synchronized Agent state and client-side ✅ SQL storage (sorry uncle bathrobe) ✅ Durable Execution of third party @YouTube APIs ✅ Human in the Loop ✅ Client + Server RPC https://t.co/mymXSnrXgu - **grinich** (2025-05-09): https://x.com/grinich/status/1920636351139230043 - imagine losing to jira but boasting that you save a couple million dollars per year dhh shares all this virtue signaling as if they are business masterminds but seems they just got smoked by linear, asana, monday, and even atlassian does anyone actually use basecamp? i literally can’t thing of a single company - **yagiznizipli** (2025-05-08): https://x.com/yagiznizipli/status/1920584336945889707 - After 2 weeks of hard work, v8 fast api support is landing soon! I'll write a blog post about the impact of this change to the @Cloudflare infrastructure. https://t.co/fGeNm19L4r - **jbarro** (2025-05-08): https://x.com/jbarro/status/1920532431431409887 - Chicago Pope, Tuesdays on NBC - **southpolesteve** (2025-05-08): https://x.com/southpolesteve/status/1920516778725105727 - I need MCP servers for @nytcooking and @Instacart and all my problems will have finally been solved - **iannuttall** (2025-05-08): https://x.com/iannuttall/status/1920509739546325375 - A couple of trolls saying "lol $25 for an MCP boilerplate I could vibe code in 5 minutes for free" if your hourly rate is $100 you've got 15 minutes MAX before it makes more sense to just buy my boilerplate ;) here are the docs, go for it: - cloudflare starter kit here: https://t.co/HauIhzzAl2 - stripe agent here: https://t.co/9YALZifRMz - **CloudflareTV** (2025-05-08): https://x.com/CloudflareTV/status/1920505262412038172 - Live in 15 min: Brendan Irvine-Broque (@irvinebroque) and special guests from @Wix look into the architecture behind the Wix MCP server. Learn how the server was built and explore how Wix Sites leverages Cloudflare's Workers for Platforms to handle backend deployment and hosting at scale. Check it out >> https://t.co/xxDUjf7k7W @CloudflareDev #CloudflareTV #MCP #CloudflareWorkers - **brian_scanlan** (2025-05-08): https://x.com/brian_scanlan/status/1920505078831530338 - Random data dump - Intercom now does 150k RPS HTTPS requests and 50k async worker requests at daily peak. Also 1 million RPS on PlanetScale! - **elithrar** (2025-05-08): https://x.com/elithrar/status/1920503336949686709 - FYI: you can just use a DB driver (like mysql2!) on @cloudflare Workers + connect to @PlanetScale if you want to build production-grade apps on your favorite RDBMS: https://t.co/ES5NS6kHdx - **grinich** (2025-05-08): https://x.com/grinich/status/1920490120169636115 - Are you coming to MCP Night? @irvinebroque will be discussing @Cloudflare support for MCP servers ⚡ RSVP here: https://t.co/yWABhuuTs5 https://t.co/hTeS4VBNpP - **willahmed** (2025-05-08): https://x.com/willahmed/status/1920486427176898599 - ANNOUNCING THE ALL-NEW WHOOP I’m proud to introduce the most ambitious leap forward in our technology ever. Two new hardware devices: WHOOP 5.0 & MG. 14 Day Battery Life. Healthspan. Heart Screener with FDA clearance. Blood Pressure Insights. New bands and apparel. And more 👇🏼 https://t.co/OyfIcDbUM2 - **joshm** (2025-05-08): https://x.com/joshm/status/1920436049882325016 - The secret about Fidji Simo is she’s one of the kindest, most generous people in tech. Yes her resume is as impressive as people say. But talk to folks behind-the-scenes and they will all tell you: Fidji is a gem of a human who touches many. Bravo OpenAI. - **RedwoodJS** (2025-05-08): https://x.com/RedwoodJS/status/1920383020348928305 - RedwoodSDK is a React framework for Cloudflare. It starts off a Vite plugin that gives you access to React server components, server-side-rendering, and an emulation environment for Cloudflare: You get database, storage, and queues that are identical to production. One package, no setup, everything you need to build software. "It's MAMP for React and TypeScript." - **fidjissimo** (2025-05-08): https://x.com/fidjissimo/status/1920376510172504394 - It will be amazing to work with @kevinweil again! https://t.co/VDTrpjw4a1 - **btaylor** (2025-05-08): https://x.com/btaylor/status/1920349717764190588 - Sam and I are so excited to welcome Fidji as CEO of Applications at OpenAI. Fidji is one of the most remarkable leaders I have ever had the privilege of working with, and I have personally seen her commitment to our mission in the boardroom over the past year. With Fidji joining Sam’s leadership team, Sam will be able to spend even more of his time scaling Research and Compute - the heart of our mission to build maximally beneficial AGI. I have never felt more optimistic about the future of our mission. - **fidjissimo** (2025-05-08): https://x.com/fidjissimo/status/1920345706663157979 - So excited to be joining @openai and contributing to its mission. Thank you @sama for the opportunity- it will be such a privilege to work with such a talented team on one of the most important and ambitious endeavors in history. I’ll remain CEO of @Instacart for the next few months and Chair of the board after that; I’m so grateful to my Instacart team for the amazing ride we had together, and I look forward to supporting the next CEO during the transition. - **gdb** (2025-05-08): https://x.com/gdb/status/1920344903466529193 - Welcome @fidjissimo! Fidji has been an amazing friend and colleague, with unique insights and advice on OpenAI. I'm super excited to work with her to deliver AGI that benefits all of humanity. https://t.co/uRul97VAB0 - **amir** (2025-05-08): https://x.com/amir/status/1920342069521829970 - breaking: instacart CEO fidji simo to run much of OpenAI soon more to come... https://t.co/IxBviaux2f - **sama** (2025-05-08): https://x.com/sama/status/1920341429655634024 - so excited that @fidjissimo is joining openai in a new role: ceo of applications, reporting to me. i'll remain ceo of openai, but in this new configuration i'll be able to increase my focus on research, compute, and safety. these are critical as we approach superintelligence. - **lizziepika** (2025-05-08): https://x.com/lizziepika/status/1920272270083150218 - Chat w/ your @cloudflaredev D1 database in @windsurf_ai by connecting them via MCP server! - help Windsurf better understand your database - make your life easier as a dev: chat w/ your data from within the IDE https://t.co/zKuhUFtk6c - **rohinlohe** (2025-05-07): https://x.com/rohinlohe/status/1920265170032443835 - .@Lutra_AI is the most impressive agentic client I've used, and they've leaned into supporting MCP. MCP tools can emit large JSON blobs, often exceeding context windows & causing client-side failures. Lutra does lots of interesting work to avoid that & stitch workflows together https://t.co/pHyhdIhIXI - **korinne_dev** (2025-05-07): https://x.com/korinne_dev/status/1920264431902015639 - We're fixing a bunch in our Cloudflare Terraform provider. Some updates: - v5.4 of the provider is out! - Major issues (like resources saying they need to be update/recreated without a change) are fixed: https://t.co/khad5ZdXQH https://t.co/gTqQ8UM1oB - **mikeyk** (2025-05-07): https://x.com/mikeyk/status/1920211404847788528 - Just shipped web search to our API! We already use this in Claude Code to dynamically fetch documentation or other info, now you can build it into your apps too. https://t.co/NvVJrKUURC - **patrickc** (2025-05-07): https://x.com/patrickc/status/1920209560591347794 - Foundation models at work for the economy. https://t.co/dDL4MjmDOK - **thegautam** (2025-05-07): https://x.com/thegautam/status/1920198569308664169 - TL;DR: We built a transformer-based payments foundation model. It works. For years, Stripe has been using machine learning models trained on discrete features (BIN, zip, payment method, etc.) to improve our products for users. And these feature-by-feature efforts have worked well: +15% conversion, -30% fraud. But these models have limitations. We have to select (and therefore constrain) the features considered by the model. And each model requires task-specific training: for authorization, for fraud, for disputes, and so on. Given the learning power of generalized transformer architectures, we wondered whether an LLM-style approach could work here. It wasn’t obvious that it would—payments is like language in some ways (structural patterns similar to syntax and semantics, temporally sequential) and extremely unlike language in others (fewer distinct ‘tokens’, contextual sparsity, fewer organizing principles akin to grammatical rules). So we built a payments foundation model—a self-supervised network that learns dense, general-purpose vectors for every transaction, much like a language model embeds words. Trained on tens of billions of transactions, it distills each charge’s key signals into a single, versatile embedding. You can think of the result as a vast distribution of payments in a high-dimensional vector space. The location of each embedding captures rich data, including how different elements relate to each other. Payments that share similarities naturally cluster together: transactions from the same card issuer are positioned closer together, those from the same bank even closer, and those sharing the same email address are nearly identical. These rich embeddings make it significantly easier to spot nuanced, adversarial patterns of transactions; and to build more accurate classifiers based on both the features of an individual payment and its relationship to other payments in the sequence. Take card-testing. Over the past couple of years traditional ML approaches (engineering new features, labeling emerging attack patterns, rapidly retraining our models) have reduced card testing for users on Stripe by 80%. But the most sophisticated card testers hide novel attack patterns in the volumes of the largest companies, so they’re hard to spot with these methods. We built a classifier that ingests sequences of embeddings from the foundation model, and predicts if the traffic slice is under an attack. It leverages transformer architecture to detect subtle patterns across transaction sequences. And it does this all in real time so we can block attacks before they hit businesses. This approach improved our detection rate for card-testing attacks on large users from 59% to 97% overnight. This has an instant impact for our large users. But the real power of the foundation model is that these same embeddings can be applied across other tasks, like disputes or authorizations. Perhaps even more fundamentally, it suggests that payments have semantic meaning. Just like words in a sentence, transactions possess complex sequential dependencies and latent feature interactions that simply can’t be captured by manual feature engineering. Turns out attention was all payments needed! - **zeeg** (2025-05-07): https://x.com/zeeg/status/1920184455886746019 - Sentry 9.16.1 now works on Cloudflare Workers + Durable Objects. Still some kinks we gotta iron out, but give it a shot! https://t.co/c0KBIZutPW - **samgoodwin89** (2025-05-07): https://x.com/samgoodwin89/status/1920179500383420505 - You can now bind a Worker to itself in Alchemy! https://t.co/egsmuQ8ffZ - **jeff_weinstein** (2025-05-07): https://x.com/jeff_weinstein/status/1920172346859794872 - Here's what's new at @stripe: https://t.co/AJ4mfIGPc3 - **grinich** (2025-05-07): https://x.com/grinich/status/1920170648191840314 - Are you coming to MCP Night? @zeeg will be demoing the @getsentry MCP server ⚡️ RSVP here: https://t.co/yWABhuuTs5 https://t.co/tIV6NSkFvS - **DrNiravPandya** (2025-05-07): https://x.com/DrNiravPandya/status/1920159122835825081 - A grade 1 hamstring strain is positive news for Warriors fans and #DubNation . A grade 1 sprain = microscopic tearing which is much better than the more significant tearing that occurs with a grade 2 or 3. How he responds to treatment, the MRI specifics (location, muscles involved, etc) will likely determine his RTP with the medical staff. - **willahmed** (2025-05-07): https://x.com/willahmed/status/1920147843861401923 - 😉 https://t.co/GQop75KcKY - **ritakozlov** (2025-05-07): https://x.com/ritakozlov/status/1920140369263575261 - sometimes simpler is better! excited for the new the @cloudflaredev workers getting started flow glow-up ✨ https://t.co/fi3wd7LyZO - **BraydenWilmoth** (2025-05-07): https://x.com/BraydenWilmoth/status/1920128741566754897 - We put up 5 pull requests for optimizations in the dashboard yesterday all thanks to your responses. You may see some of them live as early as today. Let's do this again next week. - **KNBR** (2025-05-07): https://x.com/KNBR/status/1919989187433967661 - BUDDY HIELD: "I'm Batman today. I saved the day. [Jimmy]'s still Robin." JIMMY BUTLER III: "You had another good game man! Except for that stupid thing you did at the beginning of the game when you didn't have on the right shorts. That's not real Robin and/or Batman like." https://t.co/1nY9eXLiGZ - **anthonyVslater** (2025-05-07): https://x.com/anthonyVslater/status/1919976756653543589 - Here’s Draymond Green’s perspective on the Steph Curry injury — said Curry was “deflated” — and the team’s response without him “Robin turned into Batman and what do you call Buddy? Alfred turned into Robin.” https://t.co/86BC7OzmE0 - **DrNiravPandya** (2025-05-07): https://x.com/DrNiravPandya/status/1919969612914991475 - In Rick Celebrini we trust. #DubNation - **DrNiravPandya** (2025-05-07): https://x.com/DrNiravPandya/status/1919949912600674375 - Quick halftime video for #DubNation on the left hamstring strain that was reported by the team that Stephen Curry suffered. There are multiple factors that determine the return to play timeline that the medical staff will determine based on exam, imaging, and risk of re-injury. - **clerk** (2025-05-07): https://x.com/clerk/status/1919927851140448360 - Introducing Clerk Billing The easiest subscription billing for B2B and B2C SaaS. No webhooks, no UI to build, no payment integration code, no DIY authorization engine. Just enable Billing and drop a <PricingTable/> on your /pricing page! https://t.co/DEO2cVp0co - **gill_kyle** (2025-05-07): https://x.com/gill_kyle/status/1919915034374791470 - The beauty of sync engines: 📝 update on one client 🌐 HTTP POSTs to API server 🗃️ API updates the DB 📜 DB feeds changes to WAL (write ahead log) 🔁 sync engine picks up WAL and feeds data to subscribed clients 💻 long polling clients update their own local state 0ms changes. https://t.co/UfMCtlKHO3 - **fin_ai** (2025-05-06): https://x.com/fin_ai/status/1919849485255004408 - 🚨 Intercom is now ISO/IEC 42001 certified – the first global standard for AI Management Systems. We’re one of the first customer service platforms to get it. Why does it matter? This isn’t just another cert. It means our AI features, like Fin, are built with real governance, risk controls, and accountability baked in. And, it adds to a serious stack: ISO 27001, SOC 2, HIPAA, and more – plus alignment with the EU AI Act. - **kentcdodds** (2025-05-06): https://x.com/kentcdodds/status/1919803894064152648 - Who's coming to MCP Dev Summit in San Francisco in a couple weeks!? https://t.co/c7bRITh3qt - **WHOOP** (2025-05-06): https://x.com/WHOOP/status/1919776772012331131 - Change is coming — and it’s closer than you think. Mark your calendar for 10 AM ET on May 8 👀 https://t.co/KrRAwBIotA - **willahmed** (2025-05-06): https://x.com/willahmed/status/1919776621491585467 - My open letter in today’s New York Times, Wall Street Journal, and London Times: TAKE BACK YOUR HEALTH. Most people wait – for a doctor’s visit, a diagnosis, a warning sign. They wait until something’s wrong to do what’s right. But we believe that the best things are built – and health is no exception. We want to sleep better, train better, and live better – not because we’re told to, but because we choose to. For over a decade, we’ve worked with the world’s top performers – researching what it takes to win championships, break records, and push the edge of human potential. We’ve learned that when data meets discipline, the results are extraordinary. Now that same 24/7 data can be in your hands. And it’s about to get even more powerful. You can perform better. You can live longer. You can feel more alive. You don’t have to wait for symptoms or permission. You just have to take control. Join us on 5.8.25. - **BraydenWilmoth** (2025-05-06): https://x.com/BraydenWilmoth/status/1919759276987252952 - Tell me something that could be improved in the Cloudflare dashboard. I'm feeling like fixing some issues live today. should this become a recurring "tell me tuesdays"? 😅 - **raw_works** (2025-05-06): https://x.com/raw_works/status/1919748827289968710 - serious props to @glenmaddern @irvinebroque @dinasaur_404 of @CloudflareDev - i literally fed their readme to cursor's agent and had a working remote MCP server with GH oauth in under 30 minutes. the only manual part (aka human work) was generating new GH oauth tokens, and running the server in a separate terminal so that the agent can test being the client. - **jeff_weinstein** (2025-05-06): https://x.com/jeff_weinstein/status/1919743703037464859 - a glimpse into the ai commerce future https://t.co/DVoeQQGT01 - **dok2001** (2025-05-06): https://x.com/dok2001/status/1919734470703267975 - Has anyone bootstrapped a vibe coding platform using a vibe coding platform? - **iannuttall** (2025-05-06): https://x.com/iannuttall/status/1919681466897666350 - Testing out a paid MCP server in @cursor_ai with auth. So far: - Server was a breeze to setup with @Cloudflare - Auth works well and opens the Github/Google oauth in a browser, once authorised the tools are available - Adding payments with Stripe from the example shared by @jeff_weinstein was very easy - I can get the agent to show the checkout link as plain text, not clickable I'd love to be able to render a button the user can click to go to checkout or to redirect them to checkout like I did to auth the server (cc @ericzakariasson) - **akothari** (2025-05-06): https://x.com/akothari/status/1919617803423760805 - After 12 years in the Mission neighborhood, we're excited to move @NotionHQ offices to the heart of downtown SF on Market St. And we got the warmest welcome from none other than Mayor @DanielLurie and @NedSegal. We are long San Francisco. Let's go! https://t.co/23GaE8rVnr - **KevinOConnor** (2025-05-06): https://x.com/KevinOConnor/status/1919609881637589391 - OKC deserved to lose. Horribly managed the end of that game. So dumb to foul up 3 when Jokic was OFF THE FLOOR AND THE NUGGETS HAD NO TIMEOUTS. - **dinasaur_404** (2025-05-05): https://x.com/dinasaur_404/status/1919516640258957759 - this McpAgent will be hibernating (on vacation) for the next week. if you need anything, reach out to 1800-mcp@cloudflare.com and the team will help you out 😌 hope to see some new remote MCP servers when i'm back 🧡 - **charl_dot_dev** (2025-05-05): https://x.com/charl_dot_dev/status/1919508840107692248 - Working on a decent MCP client part 2: - Chat UI based on @shadcn and Chat SDK by Vercel - Uses Workers AI which is basically free (llama-3.3-70b-instruct-fp8-fast, @michellechen dm'd some better models ❤️) - Multiple remote servers - TODO: Render UI components for input - TODO: OSS https://t.co/DN8GPwrqRi - **korinne_dev** (2025-05-05): https://x.com/korinne_dev/status/1919464174763364637 - We want your feedback! Here's a public proposal for supporting _routes.json config files in Workers with static assets: https://t.co/qthC7xjjNG The goal is to make it simpler to specify which routes invoke your Worker, as well as ease migrations from Pages. - **jachands** (2025-05-05): https://x.com/jachands/status/1919449617046872320 - Want to use a monorepo for Cloudflare Workers but not sure where to start? Starting today, you can setup a monorepo with everything needed to build and deploy multiple Workers in 15 seconds: npm create workers-monorepo@latest https://t.co/lr5NzJVk9c - **alanaagoyal** (2025-05-05): https://x.com/alanaagoyal/status/1919442680184242431 - custom gpts walked so mcp servers could run - **mattsilx** (2025-05-05): https://x.com/mattsilx/status/1919431398022164494 - Same React code, 5x times faster on @tan_stack Start vs Nextjs 15. Feels unfair tbh. Client-first approach shines. Kudos to @tannerlinsley for building not only faster, but a way more delightful DX. https://t.co/NQehFQYZml - **dinasaur_404** (2025-05-05): https://x.com/dinasaur_404/status/1919427348371259678 - Webflow Cloud — powered by @Cloudflare Workers for Platforms 🧡 https://t.co/lEyhnMMIq9 - **RedwoodJS** (2025-05-05): https://x.com/RedwoodJS/status/1919396786310922634 - RedwoodSDK gives you control over every byte over the wire. Alexander on Discord, describes how he created a SPA for some routes, and a static (no javascript) document for others. cc: @wesbos https://t.co/QBxEbnWjA3 - **mcpsummit** (2025-05-05): https://x.com/mcpsummit/status/1919388584085262827 - We're thrilled to have Dina Kozlov @dinasaur_404, Sr. Product Manager at @Cloudflare, on the speaker lineup at the #MCPDevSummit! Don't miss Dina's session on Building and Deploying Remote #MCP Servers. Click to snag a spot ➡️ https://t.co/zysOsADz6b #AI #LLM #ModelContextProtocol - **charl_dot_dev** (2025-05-05): https://x.com/charl_dot_dev/status/1919328844860866761 - Working on a decent MCP client using workers AI (No OpenAI), managing MCP servers and returning the results Using the cloudflare agents starter as a base https://t.co/HlrElEaHbM - **ccheever** (2025-05-05): https://x.com/ccheever/status/1919315995359101409 - The way that distribution of software is constrained on mobile chokes out so much valuable stuff that could be happening. This is why there are still 10x more web developers than mobile developers even though most people are on their phones more than half the time now. - **warriors** (2025-05-05): https://x.com/warriors/status/1919242769203118571 - "SURPRISE!" https://t.co/8iWj4tKVKM - **dok2001** (2025-05-05): https://x.com/dok2001/status/1919225108020826251 - @benallfree Whoa https://t.co/JPzpuaMf72 - **benallfree** (2025-05-05): https://x.com/benallfree/status/1919224851409379786 - BOOYA!!! ~~Persistent file storage for @Cloudflare Durable Objects~~ This is a real filesystem and this is a real terminal and these are real files. Behind the scenes: I ported the Rust-based chunked SQLite FUSE strategy from yesterday. The DO uses its builtin `ctx .storage.sql` API and does what the Rust implementation did with almost 100% parity. Then, I used xterm. js to create a frontend. The worker is running @honojs and as you can see it calls `stub.listDir(path)` which I exposed publicly since I want to expose the DO's filesystem calls to the worker. This is live and working, you can go upload files and browse directories. Everyone will see what you upload so please be careful and respectful. @MikeNomitch_CF @dok2001 @threepointone @rickyrobinett @maietta - **benallfree** (2025-05-04): https://x.com/benallfree/status/1919179006949326971 - Alright, we've got a first port of the FuseFS chunked SQLite backstore from Rust into Typescript. The interface looks very similar. Now you can do `this .ctx.storage.fs.writeFile(...)` and...it should work? Next step, create a worker harness that is basically a little filesystem editor. Upload, download, browse.. why not. - **benallfree** (2025-05-04): https://x.com/benallfree/status/1919160296255684783 - Now that I've got a SQLite persistent storage POC for @Cloudflare DO and containers, I took a nap and woke up realizing that I can do an even better POC in userland code. I suggested earlier that DO should have a Filesystem API. I can build it without needing any special access to CF systems. Here's the plan today: * Create a package called `dofs`, the persistent filesystem storage API extension for Durable Objects. * `dofs` has two parts: (a) `ctx .storage . fs`, and (b) a REST or WebSocket API into the same. The API will be FUSE-like so it's easy to map FUSE client calls to the API. * Build a new dofs provider in the existing Rust code to make a dofs FUSE client. The client connects to the dofs endpoint (probably https with a secret key or something) and mounts the FUSE volume ready set go! - **BraydenWilmoth** (2025-05-04): https://x.com/BraydenWilmoth/status/1919063660690870474 - The ability to provide great feedback is a superpower. Anyone can vocalize an opinion, but few can articulate quality alternate solutions. Don’t just say there is a problem, suggest how you would fix it + why. Provide references. Explain. Learn this & make positive changes. - **jeff_weinstein** (2025-05-04): https://x.com/jeff_weinstein/status/1919039241868468349 - we’re hiring a mcp-focused product engineer at @stripe to build out client and server support. you will work directly with some of the fastest growing mcp tools and applications to make cool stuff. there’s no job listing (soon!) but feel free to just email jweinstein@stripe.com. - **yusukebe** (2025-05-04): https://x.com/yusukebe/status/1918948569727582539 - Super minimal web framework (or router?) using URLPattern (not Hono!) https://t.co/URJXZ3Fnp9 - **stevekaliski** (2025-05-03): https://x.com/stevekaliski/status/1918775032601276596 - The amazing thing about MCPs being URLs is that installation can be from the share sheet from any other app that renders urls (all of them…) Share to {MCP Client} - **benallfree** (2025-05-03): https://x.com/benallfree/status/1918752661941473372 - Proof of life! Rust-based in-memory FUSE mount running @pocketbase! I've got a decent provider pattern that can use an in-memory filesystem for local testing, and once I have it completely working I'll implement a local SQLite backstore. Then I'll implement something like a REST API that maps more or less exactly to the provider I've created, which itself is a streamlined FUSE trait. With that, can implement any storage provider behind HTTP. FUSE over HTTP. What should we call it... @MikeNomitch_CF we are on our way. @Cloudflare @dok2001 @rickyrobinett - **kentcdodds** (2025-05-03): https://x.com/kentcdodds/status/1918694035793076369 - Flights and hotel booked. I'll be in SF May 22-24 for MCP Dev Summit. https://t.co/EPbcumaDHU - **jeff_weinstein** (2025-05-03): https://x.com/jeff_weinstein/status/1918687153791586669 - request for hack: mcp-business-in-a-box - an ai that auto-creates an mcp based on your api and docs - connects to your @stripe - understands your business model - instruments your mcp with stripe mcp payments - deploys it to @cloudflare (or other preferred mcp hosting platform) - **NBCSWarriors** (2025-05-03): https://x.com/NBCSWarriors/status/1918523031565087218 - Same, Canon, same 😭 https://t.co/bnVHuidg8S - **jeff_weinstein** (2025-05-03): https://x.com/jeff_weinstein/status/1918484103277756875 - Just off the phone with a growth person from a major internet company: ~"We're seeing much higher conversion rates from prospective users coming from organic LLM traffic versus organic search." - **townbrad** (2025-05-02): https://x.com/townbrad/status/1918446604467519727 - My favorite Gregg Popovich story is personal. Hopefully he won't hold it against me for telling it. Days after the San Antonio newspaper I worked for folded in 1993, Golden State assistant Pop called to tell me the Spurs were being sold and he was coming back as GM. Me: "Pop, I really appreciate it, but I have no way to publish that." Pop: "I know. I'm hoping you can use that information to get a job." - **anthonyVslater** (2025-05-02): https://x.com/anthonyVslater/status/1918445428737228919 - Steve Kerr with a statement on Gregg Popovich’s retirement from coaching “Pop’s gonna kill me for wearing this shirt. He’d call me a hapless rube.” https://t.co/iE3Ez9MNoE - **jachands** (2025-05-02): https://x.com/jachands/status/1918437234417475997 - Want to create your own Workers monorepo? I created at template to make it easier! https://t.co/74FJDPSRBK - **wemby** (2025-05-02): https://x.com/wemby/status/1918408239508488378 - 29 years. Coach, thank you for your wisdom, for your leadership, for the culture you created… But most importantly for being a great and inspiring person. It was an honor to be a part of those 29 years🙏🏽 Wishing you the best on your new chapter. 🐐 - **kentcdodds** (2025-05-02): https://x.com/kentcdodds/status/1918400523113517099 - Ok, where are MCP people hanging out? I need to find the community. - **dsp_** (2025-05-02): https://x.com/dsp_/status/1918387578845503595 - Thanks to the help of Oauth authors at Okta, Microsoft, AWS and MIT (Justin) https://t.co/yOUfvWbg3D - **_irtefa** (2025-05-02): https://x.com/_irtefa/status/1918372673014637005 - Moved to America for school. Met my wife. Built a company. Became a dad. Today, I became a citizen. Hard to put into words. But this country changed my life. Proud to be an American. https://t.co/7Of9xc5jWd - **thdxr** (2025-05-02): https://x.com/thdxr/status/1918356023032631724 - we take MRR for granted most businesses do not really have recurring revenue. have to push every day for the dollars to keep coming in it is so much harder - idk how people do it - **sentry** (2025-05-02): https://x.com/sentry/status/1918352840957018597 - Yesterday we spent some time with our friends at @Cloudflare and @AnthropicAI talking about all things MCP. In the session, @Codydearkland showed everyone how easy it is to get Sentry's remote MCP server up and running. Check out below how easy it is to get going 👇 https://t.co/YkaGWMKqiu - **AnthropicAI** (2025-05-02): https://x.com/AnthropicAI/status/1918346817378431022 - Web search is available worldwide for all paid plans. For everyday tasks, Claude runs quick searches. For more complex questions, it explores multiple sources, including Google Workspace. https://t.co/wNAlSrytvY - **0xOSN** (2025-05-02): https://x.com/0xOSN/status/1918344276108628401 - Just found the Cloudflare MCP servers repo — geeking out on how it hooks LLMs into Cloudflare services via SSE. Now I can natural-language debug Workers, spin up containers, and even tweak DNS from an MCP client. Nerdy heaven. 🚀 SSE URLs ready for instant integration. - **BraydenWilmoth** (2025-05-02): https://x.com/BraydenWilmoth/status/1918339457771614231 - Day 4: Chat w/ Uploads Showcasing more of how to build a full app with Workers, DO's and even R2 asset uploading. This Slack clone allows you to invite users, create channels, send messages and share files. - DO for Auth (1) - DO per Channel (N) - R2 uploads 👇 Github repo https://t.co/JBhCsCSzpR - **ritakozlov** (2025-05-02): https://x.com/ritakozlov/status/1918336624909222197 - this could be you https://t.co/dpJA9xVuj9 - **ritakozlov** (2025-05-02): https://x.com/ritakozlov/status/1918332073179283866 - when weekly demos are happening but you’re on a plane just watching chat go nuts 😭 https://t.co/p2yiPinx65 - **craigsdennis** (2025-05-02): https://x.com/craigsdennis/status/1918329142262808972 - 🛹 On the latest Dry Run, we had @kentcdodds show up and do kickflips and other tricks like building an MCP server from scratch. Come hang out! https://t.co/Lf2Kj9N0NV - **jeff_weinstein** (2025-05-02): https://x.com/jeff_weinstein/status/1918313735292359021 - hope to see many of you next week for ⁦@stripe⁩ sessions in san francisco. (we are shipping several dozen new things and major improvements.) message if you want a free last minute ticket! https://t.co/rkP170OTGF - **CloudflareDev** (2025-05-02): https://x.com/CloudflareDev/status/1918296415588982949 - Want an example of a real, production-grade agent — one that's powering @knocklabs - ? code here: https://t.co/nbBYZ4kGiE https://t.co/4bpMhKLZ4C - **thdxr** (2025-05-02): https://x.com/thdxr/status/1918275401484296703 - we are making our cloudflare support good this month - **harshil1712** (2025-05-02): https://x.com/harshil1712/status/1918266997822902747 - After MCP demos yesterday, and @kentcdodds's workshop, I am gonna try to build one. Join me today at 3 PM CEST for a live stream where I try to build an MCP server to manage my @CloudflareDev account. YT: https://t.co/fuL9RYBSzW Twitch: https://t.co/xfaUF66LTV - **growth_eng** (2025-05-02): https://x.com/growth_eng/status/1918162128361701403 - many saas apps will be headless in the near future, due to technologies like mcp. no one wants to log in to 10 different platforms, we're just conditioned to. - **KevinOConnor** (2025-05-02): https://x.com/KevinOConnor/status/1918133039172841873 - I know Brunson fried Ausar Thompson to win the game. But I'm still baffled why JB Bickerstaff subbed out Ausar with 3:50 left. Thompson changed the whole game when he subbed in at 9:30. Pistons were DOWN 11 when he came in and UP 5 when he came out. Awful decision to bench him. - **kentcdodds** (2025-05-02): https://x.com/kentcdodds/status/1918115878765293879 - @joelhooks Yep. Normal people words are good - **wesbos** (2025-05-02): https://x.com/wesbos/status/1918113634166731248 - Yagiz added the feature faster than I spent debugging it 😆 https://t.co/vnQ2DjKA99 - **pulsemcp** (2025-05-02): https://x.com/pulsemcp/status/1918111705953124377 - (5/7) The @CloudflareDev team has really outdone themselves. Not only did they launch 13 (yes, 13) remote MCP servers today; they also led the charge partnering with a slew of ~10 other companies to get their own live 13 servers: https://t.co/JoXSx7joa0 Demo day: https://t.co/FIoEfL3oly - **pulsemcp** (2025-05-02): https://x.com/pulsemcp/status/1918111707723120793 - (6/7) @linear came in with their remote MCP server today. MCP users love managing issues and repositories with their LLM's - after all, the GitHub MCP server is the most popular of all time. We expect Linear to follow suit https://t.co/68qPeN0J9R - **pulsemcp** (2025-05-02): https://x.com/pulsemcp/status/1918111702463525117 - (3/7) GitMCP by @idosal1 and @liadyosef has been a standout offering, bringing up-to-date developer documentation to MCP clients by way of remote URLs. Now with Claude's remote MCP support, it'll be easier than ever to have a chat with some docs. https://t.co/HLRzaAeCTZ - **pulsemcp** (2025-05-02): https://x.com/pulsemcp/status/1918111697598091387 - (1/7) Announcing 📢 Remote MCP server listings in our server directory! We crawl and catalog these from around the internet so you don't have to fish for them. There's 30+ already, and here are some of the standouts... 🧵 https://t.co/FjWjkwEg8Q - **pulsemcp** (2025-05-02): https://x.com/pulsemcp/status/1918111699758137833 - (2/7) Supermemory is a personal knowledge platform by @DhravyaShah. Collect, organize, and recall information from various sources with end-to-end encryption and optional self-hosting. https://t.co/9Hr0zGCYVu - **yagiznizipli** (2025-05-02): https://x.com/yagiznizipli/status/1918109900061679734 - Sorry for the pain @wesbos. @Cloudflare workers now support navigator.language! https://t.co/qh5TGy2uxc https://t.co/pYbX7L4wZ0 - **joelhooks** (2025-05-02): https://x.com/joelhooks/status/1918105885924311447 - reframing MCP as Integrations is going to accelerate mass adoption. Love it. https://t.co/udFAAPoz3h - **semicognitive** (2025-05-02): https://x.com/semicognitive/status/1918105083369152620 - MCP may be the primitive that gets us to CLUI, where Claude/ChatGPT is the new terminal. CLUI is the beautiful concept layed out here https://t.co/J5ZkQuEApY by @Replit @AMasad @TylerAngert https://t.co/rMwOiwCeNN - **glenmaddern** (2025-05-02): https://x.com/glenmaddern/status/1918094382290616402 - MCP only running in desktop clients is like the iPhone pre-App Store. Today is a big day. https://t.co/4eKULq6nb2 - **TziokasV** (2025-05-02): https://x.com/TziokasV/status/1918093236163494203 - It's very likely this demo becomes viral or at least gets imprinted in many people's mind thanks to the "aha moment" when the @stripe engineer shows how to monetize an MCP. It's even more 🤯when Claude counter offers to provide the service for free. You can actually see a very near future where this turns into a marketplace of MCPs with micropayments or even an MCP ranking system (a la google ads). - **gakonst** (2025-05-01): https://x.com/gakonst/status/1918091667275424140 - I'm pretty sold on paid MCPs as a way for backend devs to monetize w/o front-end engineering. How does AI search for the right MCP / set of tools, without having the user to opt-in to specific ones? https://t.co/1THPYn5816 - **lizziepika** (2025-05-01): https://x.com/lizziepika/status/1918090724937220201 - Built a remote MCP server (authless) on @CloudflareDev using Browser Rendering and Workers AI that scrapes + answers Qs about input URL content Connected it to @AnthropicAI Claude Desktop, giving it the ability to answer qs about given my personal site (or other sites)🧵 https://t.co/KtiuBApHCh - **kentcdodds** (2025-05-01): https://x.com/kentcdodds/status/1918086373812388145 - I appreciate that @AnthropicAI is taking the "Open Web" approach rather than the "Walled Garden" approach to integrations (https://t.co/2Y7c2YqDBn). Nobody is the gatekeeper to this new user interaction model and I think that's awesome. https://t.co/ElG7iinyyz - **TobinSouth** (2025-05-01): https://x.com/TobinSouth/status/1918075770267639853 - We're going to look back and see this (not GPTs) as the starting bell on the real marketplace for agent integrations and real consumer traction https://t.co/oiOEMnsQYO - **EastlondonDev** (2025-05-01): https://x.com/EastlondonDev/status/1918065397355491809 - We are making it easy to add rules from all your package.json dependencies to @cursor_ai @windsurf_ai @roocode @cline Claude code, codex and more https://t.co/mOPbdx5ORY https://t.co/JRYBWGBQcD - **aquariusacquah** (2025-05-01): https://x.com/aquariusacquah/status/1918061086219354220 - Born too late to explore earth Born too early to explore space Born just in time to pay one vendor to ETL salesforce data into my data warehouse and reverse ETL that same data back into salesforce for an annual contract value of 1.5 million dollars https://t.co/ueW7k4crqn - **davidhoang** (2025-05-01): https://x.com/davidhoang/status/1918054559668404255 - I am very excited about this not just for Atlassian, but the entire software ecosystem. Software interoperability has been this challenge in the B2B space. With MCP, we are at least working on a shared protocol that brings down the walled garden more. - **davidhoang** (2025-05-01): https://x.com/davidhoang/status/1918053756257837379 - Excited to share @Atlassian's Remote Model Context Protocol (MCP) Service is available in beta. Now you can connect Jira and Confluence directly wherever you use @AnthropicAI 's Claude; hosted on @Cloudflare infrastructure. https://t.co/17vEX0dE0T - **jachands** (2025-05-01): https://x.com/jachands/status/1918050795213447642 - The mcp-server-cloudflare repo is such a great example of why monorepos are great: 16 Workers deployed from one repo with shared packages to help ensure consistency across servers: https://t.co/Va3CYnV95d - **rickyrobinett** (2025-05-01): https://x.com/rickyrobinett/status/1918049664466862349 - All this vibe coding is making me hungry... So I built a remote MCP Server with @CloudflareDev that lets me order 🍕 directly from within Claude Code. https://t.co/LY3QzQNkVI - **alexalbert__** (2025-05-01): https://x.com/alexalbert__/status/1918047745790914772 - Not sure if some folks are fully realizing what we launched today so to make it more explicit: You can bring any custom MCP server into claude dot ai now. All you need is to host it somewhere (remote MCP) and provide the URL link to add it in. - **jeff_weinstein** (2025-05-01): https://x.com/jeff_weinstein/status/1918046167394598969 - MCP-only software is coming. https://t.co/DThcrXZMcp - **destraynor** (2025-05-01): https://x.com/destraynor/status/1918043352991146470 - 🆕Fin now supports MCP as a client, and Intercom supports MCP as a server. This little example shows Fin connecting to @stripe's MCP server letting Fin list products, prices, and even create payment links. This lets Fin resolve customer questions better, faster, and cheaper https://t.co/cqLoSvkq2z - **jeff_weinstein** (2025-05-01): https://x.com/jeff_weinstein/status/1918036046866157809 - @prosumerworks @stripe @Cloudflare With @stripe, your MCP server can accept ~any payment method, including stablecoin! https://t.co/cWnQFNZIPh - **jeff_weinstein** (2025-05-01): https://x.com/jeff_weinstein/status/1918035241517891908 - @irvinebroque Get started today => https://t.co/jXn7xEvoW1 - **jeff_weinstein** (2025-05-01): https://x.com/jeff_weinstein/status/1918035057383752161 - @kentcdodds @irvinebroque Here's more Kent, we'd love your feedback => https://t.co/jXn7xEvoW1 - **CloudflareDev** (2025-05-01): https://x.com/CloudflareDev/status/1918032657474048121 - Dry Run: Ep 6: Kent C. Dodds builds an MCP Server https://t.co/QJFMRHMaLl - **jeff_weinstein** (2025-05-01): https://x.com/jeff_weinstein/status/1918029261430255626 - Dearest MCP developers, You can now monetize your MCP in a few lines of code with @stripe, available today. 💸 - Bill subscriptions or usage-based - Works with any client - Supports @Cloudflare's Agents SDK (more to come) MCP, a new AI-native customer channel. Get started. ⤵️ https://t.co/CB2BQZTSpI - **darraghcurran** (2025-05-01): https://x.com/darraghcurran/status/1918023814375424479 - when you hire people you give them access to the best tools they need to do their job. The same ought to be true for AI agents - giving them the right tools for the job is becoming easier with MCP. https://t.co/MuhROdjGev - **kentcdodds** (2025-05-01): https://x.com/kentcdodds/status/1918015813040390413 - @DamiDina @irvinebroque @jeff_weinstein Here's the demo and my reaction: https://t.co/x3jmPFuknU - **Codydearkland** (2025-05-01): https://x.com/Codydearkland/status/1918012113500660051 - Me in Slack at @getsentry after finishing the MCP + @CloudflareDev + @AnthropicAI demo. https://t.co/lOoM5UZc4g - **stevekaliski** (2025-05-01): https://x.com/stevekaliski/status/1918011335368986940 - @kentcdodds @irvinebroque @jeff_weinstein Send us a DM! Would love to get ya set up :) - **jeff_weinstein** (2025-05-01): https://x.com/jeff_weinstein/status/1918008619188273482 - Truly the internet is being rebuilt and the preview of it is this set of demos: https://t.co/gs6ex5hqLq - **liadyosef** (2025-05-01): https://x.com/liadyosef/status/1918006555955610002 - Wow, https://t.co/A8ATGzZiPA now natively supports GitMCP! 🚀 Check out this example with React Router docs. How to add any GitMCP integration: 1. In “Profile > Integrations” click “Add more” 2. Add any repo's GitMCP server by replacing https://t.co/m3PldkBedz with https://t.co/0kzV11CsvV in its URL - or use https://t.co/MiQZR9fYMX for *all* repos. This is amazing. Great work by @AnthropicAI . - **fin_ai** (2025-05-01): https://x.com/fin_ai/status/1918006078429671692 - AI agents get better with the more data they have access to. With MCP, Fin connects to your tools in minutes – not months. Better data = better support = better outcomes. The future of AI customer service is connected. 👉 https://t.co/Sa5kZD06LC - **kentcdodds** (2025-05-01): https://x.com/kentcdodds/status/1918002528786022542 - @irvinebroque @jeff_weinstein This demo from Mathew is one of the coolest parts of the day. I'm blown away. - **gching** (2025-05-01): https://x.com/gching/status/1918002006985478486 - @Square demoing their MCP server at @CloudflareDev MCP Demo Day is so sick, managing a storefront using AI will definitely be the future, we just need to think about the UX now Thank you Mike from Square and Emily 🤭 https://t.co/BmcrjFcrrx - **dok2001** (2025-05-01): https://x.com/dok2001/status/1918000721745580266 - Tun into @Cloudflare MCP Demo Day! https://t.co/WQJDRGatAA - **swyx** (2025-05-01): https://x.com/swyx/status/1917999320055582948 - Claude starting to flex remote MCPs and adding them to 45 minute-long Claude Deep Research. nice. theres the developers who complain that mcps dont offer anything over rest apis then theres the companies that wake up one day and they're integrated to claude/every mcp client without needing to be asked bc all you need is one string - **Codydearkland** (2025-05-01): https://x.com/Codydearkland/status/1917998133096878087 - About to go live on this! You're going to watch me go from 0 to MCP implemented (or, fail VERY publicly), and then dive into pulling @getsentry context into Claude and my IDE. We'll talk building MCP on @Cloudflare, @AnthropicAI and some good workflow vibes! Lets gooooo. https://t.co/7dvatlOHQV - **pparthasarathy0** (2025-05-01): https://x.com/pparthasarathy0/status/1917991094987506044 - watching @Cloudflare's MCP demo day (featuring @stripe and other amazing companies). what a time to be alive. https://t.co/aKR7eeGEwV - **sentry** (2025-05-01): https://x.com/sentry/status/1917990482082480313 - MCP is the best way for teams to pull context from external systems into their LLMs. Building it on @Cloudflare has been a delight. - Easy to spin up and get to building - Remote MCP ❤️ - Big big scale - OAuth - Durable Objects Check out https://t.co/z4WteMBjhc for more https://t.co/8lfnMPAFTY - **JacobMGEvans** (2025-05-01): https://x.com/JacobMGEvans/status/1917989849136808155 - This is AWESOME! https://t.co/v1mpDa9DxJ - **boristane** (2025-05-01): https://x.com/boristane/status/1917988331775762639 - 13 Cloudflare MCP servers 🤯 Workers Observability is right there, alongside autorag, bindings, and so many more https://t.co/TtQQFd3oap - **mcp_host** (2025-05-01): https://x.com/mcp_host/status/1917987572480876778 - Happy MCP Demo Day to those who celebrate 🥳 https://t.co/k6pFMInnug - **craigsdennis** (2025-05-01): https://x.com/craigsdennis/status/1917987110868664366 - Come hang out!!!! https://t.co/60p0UF4hMm - **CloudflareDev** (2025-05-01): https://x.com/CloudflareDev/status/1917986456070394079 - MCP Demo Day LIVE! https://t.co/IrRvY05lEs - **kristianfreeman** (2025-05-01): https://x.com/kristianfreeman/status/1917984245928124528 - We published a new blog post today about how ten companies - Anthropic, Atlassian, Stripe, and more - shipped their own remote MCP servers on @Cloudflare 🔥 https://t.co/c5xSBOrkX6 - **dsp_** (2025-05-01): https://x.com/dsp_/status/1917982784213193068 - @ritakozlov Couldn’t do it without you all that make deploying remote MCP easy - **darraghcurran** (2025-05-01): https://x.com/darraghcurran/status/1917981849881596259 - It's been fun for our team partnering with and building on Cloudflare. MCP will massively opens up the capabilities of https://t.co/s9lTJsUwhu - imagination and creativity becomes the limiting factor. https://t.co/pHGQDYkqwq - **ritakozlov** (2025-05-01): https://x.com/ritakozlov/status/1917981417910534388 - @dsp_ huge congrats!!! can't believe how far MCP has come - **ritakozlov** (2025-05-01): https://x.com/ritakozlov/status/1917980371322589542 - have a question about @cloudflare ? ask claude. debugging an issue? ask claude. setting up dns? claude can do it all — we've added 13 cloudflare MCP server(s!) that are now live read our blog post for details, and 4 important lessons we learned along the way 👇 https://t.co/8uwE3YfZcL - **karrisaarinen** (2025-05-01): https://x.com/karrisaarinen/status/1917980275709264308 - The official @linear MCP is live! It follows the authenticated remote MCP spec and supports issues, projects, and comments via Claude, @cursor_ai, @windsurf_ai, and more. Thanks @AnthropicAI + @Cloudflare for partnering on this https://t.co/Y3xsnMTMCd - **dsp_** (2025-05-01): https://x.com/dsp_/status/1917979874314064199 - What this means is that you can bring your own Remote MCP server to https://t.co/Xkv1cxb93e https://t.co/OpNpBlEuGw - **cjc** (2025-05-01): https://x.com/cjc/status/1917979745922548072 - Linear's Remote MCP Server is now available and supported natively as a new Claude Integration and in clients like @cursor_ai and @windsurf_ai. Thanks to our partners at @AnthropicAI & @Cloudflare for helping us get this out the door! https://t.co/xxqIn0aR8s - **craigsdennis** (2025-05-01): https://x.com/craigsdennis/status/1917977526569451593 - SOOOOOOOON https://t.co/ShTkVP0nzK - **eastdakota** (2025-05-01): https://x.com/eastdakota/status/1917977196838391837 - The future is being built on @Cloudflare. https://t.co/xFOZDOcFvR - **dsp_** (2025-05-01): https://x.com/dsp_/status/1917977093528522836 - This one is big https://t.co/Xpfi8S9QuX - **Cloudflare** (2025-05-01): https://x.com/Cloudflare/status/1917977050247237770 - Cloudflare is teaming up with Anthropic, Asana, Atlassian, Block, Intercom, Linear, PayPal, Sentry, Stripe, and Webflow to launch new remote MCP servers, built on Cloudflare. https://t.co/XZBYU0AFAs - **eastdakota** (2025-05-01): https://x.com/eastdakota/status/1917976406233154028 - This is a big deal. https://t.co/QztBAsbAIa - **Cloudflare** (2025-05-01): https://x.com/Cloudflare/status/1917976344119382132 - You can now connect to Cloudflare's first publicly available remote Model Context Protocol (MCP) servers from any MCP client that supports remote servers. https://t.co/UdTdGBt1rz - **RobbieSap90** (2025-05-01): https://x.com/RobbieSap90/status/1917976268995203143 - WTF is MCP? Tune in here to learn how companies are building their MCP servers with @CloudflareDev . 🔗in thread https://t.co/6Pa15Og0pf - **ritakozlov** (2025-05-01): https://x.com/ritakozlov/status/1917976147125735438 - remote MCP is live in claude! your remote MCP server is live on @cloudflare in just a few minutes! your users can start interacting with your service directly through claude! life is good 😎 https://t.co/O2ZvqsOsH2 - **linear** (2025-05-01): https://x.com/linear/status/1917975398828134698 - Now online: Linear MCP Server Give AI models and agents access to your Linear data in a simple and secure way. https://t.co/UIPBiJk8vb - **leinwand** (2025-05-01): https://x.com/leinwand/status/1917975317546766700 - Amazing to see @webflow and @Cloudflare partner on Claude/MCP (@AnthropicAI). Go @utsengar! We can't wait to see what you build! 🤖🎉🙏 https://t.co/gIX6IHi5zv - **rickyrobinett** (2025-05-01): https://x.com/rickyrobinett/status/1917974806806352137 - I'm super hyped about MCP! Let me quickly walk you through a couple @CloudflareDev remote MCP Server templates to get you started. https://t.co/awiuDzPh6G - **alexalbert__** (2025-05-01): https://x.com/alexalbert__/status/1917973599044116576 - Two big updates to claude dot ai today: 1. Integrations using remote MCP is finally here 2. A more advanced version of Research is out Let's walk through them: - **AnthropicAI** (2025-05-01): https://x.com/AnthropicAI/status/1917972747000692919 - Today we're announcing Integrations, a new way to connect your apps and tools to Claude. We're also expanding Claude's Research capabilities with an advanced mode that searches the web, your Google Workspace, and now your Integrations too. https://t.co/zSvPBZYq43 - **kentcdodds** (2025-05-01): https://x.com/kentcdodds/status/1917971453498085727 - Looking forward to today's MCP Demo Day with @CloudflareDev and friends. I'm also going to be joining CloudflareTV to show how to build an MCP server! https://t.co/1dhiuhr18A Join us :) - **Osintly** (2025-05-01): https://x.com/Osintly/status/1917964184198823951 - We’ve just migrated our entire infrastructure from Vercel to Cloudflare ⛅ In the spirit of transparency, here’s a breakdown of why we did it, how it went, and what it means moving forward. 👇 https://t.co/kpkbSpmzgQ - **BraydenWilmoth** (2025-05-01): https://x.com/BraydenWilmoth/status/1917961344445255978 - Taking a pause today on releasing something with DO because even BETTER news... A new D1 experience is available in Cloudflare today. https://t.co/SAQQmDu6pS - **burcs** (2025-05-01): https://x.com/burcs/status/1917946901074235481 - it's not too late to be like the cool kids... come check out these mcp demos by an absolutely stacked line up https://t.co/9rJ4i25rqr - **CloudflareDev** (2025-05-01): https://x.com/CloudflareDev/status/1917946358859989448 - Today at 10AM PT / 1PM ET / 7PM CEST ⚡️ https://t.co/EtYQ2t5vcO - **jeff_weinstein** (2025-05-01): https://x.com/jeff_weinstein/status/1917940780930588909 - MCP demo day. Today. 10am PT. Livestream. Come see the future of the mcp-based internet. (And the @stripe demo, is, well, money.) https://t.co/wzUeUAend9 - **KevinOConnor** (2025-05-01): https://x.com/KevinOConnor/status/1917809038227546589 - The Timberwolves shot 7/47 from 3 tonight, the worst 3PT% (14.9%) in NBA playoff history when a team had at least 34 attempts. Minnesota missed 18-straight. For perspective, Houston shot 15.9% (7/44) in its infamous 0/27 Game 7 loss. The Wolves shot worse but pulled out the win. - **snowmaker** (2025-05-01): https://x.com/snowmaker/status/1917781471479333201 - @TimSuzman It's interesting that in a supposedly efficient market, such a big arbitrage can just lie around for so many years. - **jeff_weinstein** (2025-05-01): https://x.com/jeff_weinstein/status/1917760085172052295 - I just so happened to be walking by this (~motivational?) decision making principle poster in the @stripe office when I saw this below @patrickc tweet. "Non-consensus is good. We do not seek consensus in decision-making at Stripe. Being consensus-driven will slow us down and may hinder our ability to make progress in unconventional and innovative ways." - **ritakozlov** (2025-05-01): https://x.com/ritakozlov/status/1917745996303397154 - resume mcp! beginning of a new era https://t.co/PyVtifC86c - **byArunKai** (2025-04-30): https://x.com/byArunKai/status/1917728442662936704 - 👾 Announcing v1 of MCP Arun Kai 👾 Dear @Cloudflare , You've enabled me so much this past year. Everything I do now has cloudflare written around it. I'd like to make it official. Let me come work for you. My MCP awaits your message. https://t.co/eAbplDGjBq - **patrickc** (2025-04-30): https://x.com/patrickc/status/1917728085589516590 - At Stripe, @gaybrick always tries to disaggregate internal cases for doing something into the constituent beliefs, principles, proposal, and risks. BPPR. Can usually be done in a page. It's the most helpful way of structuring such things that I've found. - **Codydearkland** (2025-04-30): https://x.com/Codydearkland/status/1917709277122421180 - Listening to my daughters upstairs trying to learn how to rap "Alphabet Aerobics", and feeling a lot of joy about being a father. Find the special moments when they happen, and sit in them. - **fireblaze_15** (2025-04-30): https://x.com/fireblaze_15/status/1917695917697728694 - @Cloudflare R2 Object Storage Updates! 🧵 We're making improvements to our dashboard experience and here's what dropped today! ✅ All-new Settings page to manage your bucket! (1/6) https://t.co/VevJnyqlc8 - **jeff_weinstein** (2025-04-30): https://x.com/jeff_weinstein/status/1917686302855815548 - Thrilled to show you something @stripe has been building to help MCP developers. See you tomorrow (May 1) 10am PT, live, at MCP demo day => https://t.co/SKdRHLVpX3 - **jeff_weinstein** (2025-04-30): https://x.com/jeff_weinstein/status/1917684477456966130 - dear global fintech nerds, you can now create stablecoin-linked @visa cards via @stripe bridge. let your customers pay by card, worldwide, from their stablecoin balance. https://t.co/uhmXLzpukF - **grinich** (2025-04-30): https://x.com/grinich/status/1917656606042644767 - MCP Night will feature a demo by @kyliestew from @Cloudflare 🔥 Looking forward to seeing some MCP magic. RSVP to get on the list 👇 https://t.co/OHb4jvYgHv - **aidenybai** (2025-04-30): https://x.com/aidenybai/status/1917621217076330637 - you should learn how to build a Virtual DOM (the way React does UI rendering) it's easy and fun to learn. it takes <100 LoC to implement! https://t.co/sTIBbgo2OG - **BraydenWilmoth** (2025-04-30): https://x.com/BraydenWilmoth/status/1917595642630922445 - Day 3: Multiplayer Cursors Collaboration in our favorite tools has taken off & what better way to showcase than seeing where everyones cursors are on a page. Durable Objects make working with sockets so easy. ~100 lines of code but I did it for you so you can ship. 👇 Demo - **yusukebe** (2025-04-30): https://x.com/yusukebe/status/1917595570367258722 - Beautiful implementation for Remote MCP 💫 https://t.co/LyXsDBkr1h - **boristane** (2025-04-30): https://x.com/boristane/status/1917586440554160183 - I'm hiring my first role at @cloudflaredev if you want to come hack ai apps and tooling with @threepointone , @ritakozlov , @southpolesteve , @irvinebroque , @zebassembly , @elithrar , @dinasaur_404 and the rest of the team, here's THE opportunity https://t.co/qqtUWzO6r6 - **dinasaur_404** (2025-04-30): https://x.com/dinasaur_404/status/1917576300958457982 - 🚨 tomorrow (May 1st) @ 10:00 PT/1 PM ET 🎞️ MCP demos from @asana, @Atlassian, @intercom , @linear , @PayPal , @getsentry , @Square , @stripe , @webflow (and more!) will make you feel like you're living in the future ✨ trust me, you don't want to miss it. register 👇 https://t.co/ccRog7YNR9 - **dinasaur_404** (2025-04-30): https://x.com/dinasaur_404/status/1917568131255463943 - aaaaand you can use the latest version of mcp-remote to test your MCP server with the streamable HTTP endpoint 😌 https://t.co/kHy6vYbTUt https://t.co/D1B8mmqJOX - **dinasaur_404** (2025-04-30): https://x.com/dinasaur_404/status/1917566898549510633 - Now available for Cloudflare MCP servers: 🚡 Streamable HTTP transport: In just a few lines of code, upgrade your server to support both remote transport methods (streamable http + sse) 🐍 Deploy MCP servers written in Python Best part? Just ✨one-click✨ to get started! https://t.co/DEWTxl19XW - **boristane** (2025-04-30): https://x.com/boristane/status/1917408803466743846 - usage-based limits and pricing with @CloudflareDev workers analytics engine can't get any easier https://t.co/DQdKKipitW - **kentcdodds** (2025-04-29): https://x.com/kentcdodds/status/1917356814926217507 - Well would you lookie here 👀 https://t.co/xdzZQyhhlN - **goose_oss** (2025-04-29): https://x.com/goose_oss/status/1917342043011010796 - Next stop: San Francisco 🛣️✨ We’re bringing the AI Agent + MCP meetup roadshow to the Bay! Catch talks and live demos from our very own @techgirl1908, as well as speakers from @AnthropicAI, @Cloudflare, @supabase, and @stytchauth! 🔥 Come see what agents can really do. 🦾 Space is limited so RSVP now to grab your spot! https://t.co/iYs7Ju6Wvr - **EastlondonDev** (2025-04-29): https://x.com/EastlondonDev/status/1917340151237222661 - Llms.txt is great but there are some challenges: How does AI know where to get the llms.txt from for software packages ? (Your website, GitHub, your package manager) How do you get a version of the llms.txt to match the version of the software that you’re using? How do you get the correct instructions into your IDE and to your team? vibe-rules introduces a way to distribute rules for AI IDEs as part of your npm packages. Add rules to your package as a vibe-rules compatible ./llms module and export it from package.json your users can run vibe-rules install cursor/windsurf/etc And the correctly versioned rules for your package will be installed into their ide from your node module. You can define multiple rules file globs / etc. vibe-rules will take care of installing the rules into the right place AND keeping them up to date as your users change versions. - **darraghcurran** (2025-04-29): https://x.com/darraghcurran/status/1917335789337284975 - s/more/@intercom/ https://t.co/ndwct2jjN1 - **chewcodes** (2025-04-29): https://x.com/chewcodes/status/1917326428825346389 - @CloudflareDev @webflow @paypaldev @linear @getsentry @SquareDev @StripeDev @intercom This is like the Avengers for SaaSs - **michellechen** (2025-04-29): https://x.com/michellechen/status/1917319719998156894 - vibe coded a little experiment - was curious how effective it is to route requests to different models based on the complexity of the prompt. https://t.co/9MV5YBaXlc 🧵 here's what i learned: - **eastdakota** (2025-04-29): https://x.com/eastdakota/status/1917310045588496583 - It’s so strange to me how AWS defies the natural tendency of tech to make things cheaper over time and instead either holds customer prices high as their underlying costs drop (e.g., egress) or starts charging for things like cold starts creating no incentive for them to improve. https://t.co/4UvJ11S3DU - **wesbos** (2025-04-29): https://x.com/wesbos/status/1917292667597381649 - @appfactory Yeah that would be awesome! Will slide into DMs - **ritakozlov** (2025-04-29): https://x.com/ritakozlov/status/1917283240429117798 - MCP demo day coming up this thursday! learn about how @atlassian, @intercom, @linear, @paypal, @getsentry, @stripe, @webflow and of course @cloudflare are leveraging and building MCPs (deets in site or mcp on /sse). can't wait! https://t.co/RcjKCTYDYh https://t.co/FlrDtUs4eB - **fergal_reid** (2025-04-29): https://x.com/fergal_reid/status/1917276275858620900 - A new entry on our Intercom AI blog from @_ketanbhatt, about our decision to stick with @elastic as opposed to adopting a new VectorDB for Fin. tl;dr: operational knowledge is valuable, and ES will do vectors. Link below! - **appfactory** (2025-04-29): https://x.com/appfactory/status/1917274624154337293 - @wesbos Thank you! Do you have guests on Syntax? I would love to tell you more about why we've made these decisions, and what makes RedwoodSDK feel so darn good! - **zeeg** (2025-04-29): https://x.com/zeeg/status/1917270368194289864 - that logo is close enough for me https://t.co/OQTKiGXUcX - **wesbos** (2025-04-29): https://x.com/wesbos/status/1917264680524935450 - Demo comes with Passkey auth - pretty cool https://t.co/N6KJNOqQEc - **wesbos** (2025-04-29): https://x.com/wesbos/status/1917263290104963507 - Interesting pivot for Redwood - now RedwoodSDK - a React framework for Cloudflare with full support for RSC Vite Plugin, file based routing, middleware, DB migrations and tight (but not required) integration with Cloudflare D1, durable objects, queues etc.. https://t.co/s5zoGy5FSQ - **boristane** (2025-04-29): https://x.com/boristane/status/1917262766819168267 - see you there https://t.co/Ola4V2IYey https://t.co/BuUVXz6W8q - **sudo_moot** (2025-04-29): https://x.com/sudo_moot/status/1917259665328726508 - very excited for @SquareDev team to demo what our @Square MCP server can do! https://t.co/DJcftrA6Qo https://t.co/EhUUg1JoOK - **leinwand** (2025-04-29): https://x.com/leinwand/status/1917254201006027020 - Let's go! 🚀 https://t.co/URhK0aqZp2 - **ritakozlov** (2025-04-29): https://x.com/ritakozlov/status/1917249825151516748 - we had to make an mcp server for our mcp demo day! https://t.co/OrEl3pzrcM - **sentry** (2025-04-29): https://x.com/sentry/status/1917247649587347885 - We hear people are really liking this MCP thing. We too really like MCP. @Cloudflare is also pretty great - so of course we're excited to jump in on this! Come hang out and learn about all things MCP and Cloudflare! It's going to be fun! More robots! 🤖 https://t.co/h8mL4BI3AP - **burcs** (2025-04-29): https://x.com/burcs/status/1917241814836207923 - had some fun with the mcp demo day site while the animations are cool, the event is even cooler 😎 https://t.co/5OWJyHDxSy - **PeterSaulitis** (2025-04-29): https://x.com/PeterSaulitis/status/1917239928335090097 - 👀 MCP DEMO DAY 👀 https://t.co/n7HeUGYpna - **CloudflareDev** (2025-04-29): https://x.com/CloudflareDev/status/1917238218753196358 - We invited a few friends from @webflow @paypaldev @linear @getsentry @SquareDev @StripeDev @intercom (+ more) to give you a peek into the future of agentic software with MCP! Streaming Thursday (May 1) at 10 am PT. Join: https://t.co/3UGVy2MfvZ https://t.co/x55p04FEWA https://t.co/1mRyUxPqjn - **dinasaur_404** (2025-04-29): https://x.com/dinasaur_404/status/1917236219706225036 - Mark 👏 Your 👏 Calendars 👏 This Thursday, May 1st - 10AM PT / 1PM ET Join us for MCP Demo Day to see the ✨magical✨ new experiences that MCP unlocks, with demos from @Atlassian, @getsentry, @Linear, @Webflow & more! - **CloudflareDev** (2025-04-29): https://x.com/CloudflareDev/status/1917187313672597757 - SF, this Sunday: https://t.co/OH1tMwsP75 - **KNBR** (2025-04-29): https://x.com/KNBR/status/1917105597989982535 - Jimmy Butler was told that Buddy Hield thinks he's the Alfred to Steph and Jimmy's Batman and Robin. He's not having it. "NO! NO! Buddy is not in this! Buddy's the Joker or something, he's the evil man!" 😂😂😂 https://t.co/xh7L6A7SAY - **957thegame** (2025-04-29): https://x.com/957thegame/status/1917101135216972091 - “Get me on record for this: I don’t like Dillon Brooks. We are never having fun.” Jimmy Butler III on the back-and-forth with Dillon Brooks that the cameras caught https://t.co/PWBlhg9aXS - **ohnohedidnt24** (2025-04-29): https://x.com/ohnohedidnt24/status/1917098911610585180 - Jimmy Butler: "I wanted that rebound. I told Dray if you get a stop, I will get the rebound and he got the stop and I got the rebound" https://t.co/LS7C9Vbkgi - **KNBR** (2025-04-29): https://x.com/KNBR/status/1917093957634556100 - "You can't play the game worrying... I'll just go sit down if I'm just going to go out there and be cautious." Draymond Green had NO concern about getting fouled out or picking up his 2nd technical tonight. "I always walk that line. That's who I am. Habitual line stepper." https://t.co/qKzDFPnEbq - **theo** (2025-04-29): https://x.com/theo/status/1917089219790639208 - Welcome back, Sony Walkman https://t.co/YZWCWegQin https://t.co/YCgWSZEzbL - **lizziepika** (2025-04-29): https://x.com/lizziepika/status/1917086351352336892 - @irvinebroque i might cri so good - **BleacherReport** (2025-04-29): https://x.com/BleacherReport/status/1917083316001349919 - "I got my joy back as some would say." Jimmy after Warrior win 😅🙌 https://t.co/16Pb93MwV8 - **KNBR** (2025-04-29): https://x.com/KNBR/status/1917083236246720945 - Chase looking and sounding a bit more like Oracle 😏 https://t.co/4DqaC9Vm7G - **BillSimmons** (2025-04-29): https://x.com/BillSimmons/status/1917082745022648679 - Great D by Green but you gotta love the rebound by Jimmy. https://t.co/P29Egq1fhX - **agarwal** (2025-04-29): https://x.com/agarwal/status/1917081542196924541 - One of the most stressful games I’ve ever watched. I love the playoffs!! https://t.co/gc7GFcYm8y - **NoContext_GSW** (2025-04-29): https://x.com/NoContext_GSW/status/1917081238889759201 - https://t.co/wymJ77HGx8 - **NBCSWarriors** (2025-04-29): https://x.com/NBCSWarriors/status/1917081215896850698 - "NO, NO, NO. NOT ON ME." Draymond wasn't giving up a late bucket tonight 🤩 https://t.co/LtgHztfdXe - **gswchris** (2025-04-29): https://x.com/gswchris/status/1917080954205835330 - draymond disgusted they tried him lmaoooo https://t.co/BIMKAvhJAF - **TheWarriorsTalk** (2025-04-29): https://x.com/TheWarriorsTalk/status/1917080941237047454 - Jimmy Butler on his Warriors teammates: "I'd die for these guys." - **DrNiravPandya** (2025-04-29): https://x.com/DrNiravPandya/status/1917080849813803023 - Ime Udoka's main coaching skill is losing to the Warriors regardless of where he is coaching - **NBCSWarriors** (2025-04-29): https://x.com/NBCSWarriors/status/1917080186778226751 - DRAYMOND GETS THE STOP ❌ https://t.co/teYnj6UjFd - **liadyosef** (2025-04-29): https://x.com/liadyosef/status/1917060408709415055 - Cloudflare released an MCP for their docs - and it's automatically in GitMCP chat now for Cloudflare! It's pre-installed with the Cloudflare docs MCP. Check it out: https://t.co/qiZUp18zEn (and as always - there's a chat and an MCP server for any other library out there!) - **WorldWideWob** (2025-04-29): https://x.com/WorldWideWob/status/1917056908806303911 - ✅Technical Foul ✅Flagrant Foul ✅Shooting Foul ✅Loose ball Foul ✅Personal Foul ⬜️ Take Foul ⬜️ Intentional Foul ⬜️ Double Foul Draymond is out here praying the foul rosary tonight. - **liadyosef** (2025-04-29): https://x.com/liadyosef/status/1917031466866819441 - @nevikashah @itszerfl @ritakozlov @Cloudflare I was just about to offer to use https://t.co/Gk5mxSEqdF for that, but that was quick! Great job @nevikashah and the team 👏 - **zeeg** (2025-04-28): https://x.com/zeeg/status/1916984916891557990 - @irvinebroque the Class.fetch thing is very CF specific, whereas "I have a Node request object, and I need to return a Node response object" makes things a lot easier to reason about it also makes e.g. generalized middleware plausible - **zeeg** (2025-04-28): https://x.com/zeeg/status/1916984699890896919 - @irvinebroque tl;dr what I want is easy composable interfaces: route(request): Response megaRoute(request): { return mapToRoute([route]) } - **zeeg** (2025-04-28): https://x.com/zeeg/status/1916983751860797639 - @irvinebroque former (your first example) I just dont see the value in the complexity if I just want to expose a simple stateless service (e.g. a web RPC function) - **zeeg** (2025-04-28): https://x.com/zeeg/status/1916983466320924755 - @irvinebroque @samuelcolvin its just "another thing" in there that takes cognitive load JS has this problem in spades and then you add CF on top of it, and whenever anything goes wrong you're stuck debugging _every_ random thing someone else put into the project for you - **grinich** (2025-04-28): https://x.com/grinich/status/1916946294184284189 - There are now over 300 builders registered for MCP Night! 🤯 We have a few more special guests that will be announced soon... 🤫 Plus swag, food, and giveaways. Get on the list here: https://t.co/yWABhuuTs5 https://t.co/vKgz75l1DR - **nevikashah** (2025-04-28): https://x.com/nevikashah/status/1916938955821371479 - @itszerfl @ritakozlov @Cloudflare You're not babbling 🙂 we had the same thought! Here's an MCP server for our docs if you'd like to try it out: https://t.co/8SDkhVdrQw Would love your feedback! - **michellechen** (2025-04-28): https://x.com/michellechen/status/1916936805816914222 - the durable object mafia has gotten to me https://t.co/aT8tXYY2B3 - **branmcconnell** (2025-04-28): https://x.com/branmcconnell/status/1916931311253549458 - @aidenybai Have you tried @knocklabs? Knock is an all-in-one messaging/notification platform that takes in your API keys for email, SMS, push, and chat notifications @twilio, @telnyx, and @messagebird support API-driven SMS — Knock supports them all https://t.co/ghGx8KbDv4 - **elithrar** (2025-04-28): https://x.com/elithrar/status/1916914541792764319 - Love to see more great teams (like @knocklabs!) building on top of the Agents SDK! 🤝 https://t.co/GrUSm8XPF3 - **kneeovertoesguy** (2025-04-28): https://x.com/kneeovertoesguy/status/1916907713575195004 - I hope this is the most detailed deep squat thread you’ve ever read. Step 1: 25 pound plate or less is ideal for full reach. (Once COUNTERBALANCE clicks, you can be among the world’s most effective knee coaches, regardless of background. I only know from living it!) Onward… https://t.co/4MFghC8Q5U - **cjbell_** (2025-04-28): https://x.com/cjbell_/status/1916903684518060217 - It’s day one of launch week @knocklabs and we’re starting things off with a bang. Introducing Knock Agent Toolkit. Agent Toolkit exposes Knock’s APIs as tools to your AI agents so you can effortlessly build cross-channel agent to user messaging flows. https://t.co/EIoZmFet8v - **hyperbrowser** (2025-04-28): https://x.com/hyperbrowser/status/1916901537894568202 - @Stagehanddev @ycombinator https://t.co/8VKvuPW3BL - **stytchauth** (2025-04-28): https://x.com/stytchauth/status/1916901307086213158 - Catch us at the Dumb Things AI Hackathon 🚀 We’re teaming up with @Cloudflare , @replicate , and @OpenAI to build delightfully unhinged AI apps—no idea too small, no project too dumb 🤖 Expect talks, a live demo from our CTO @juliannaelamb food, drinks, and prizes 🏆 🎟️ Grab your free ticket and join us Sunday, May 7 👉 https://t.co/zAXnFrLG8f - **BraydenWilmoth** (2025-04-28): https://x.com/BraydenWilmoth/status/1916868890237309121 - Day 1: Websocket Multiplexing Consolidate sockets between client <> server to 1 but letting the server multiplex (collect/handle) all the outgoing sockets to other DO's. Client <(1)> User DO <(N)> Channel DO TLDR; Client can talk to 32,000+ socket connections via 1 socket. https://t.co/JrZPmp9rVY - **jueungraceyun** (2025-04-27): https://x.com/jueungraceyun/status/1916563724032471094 - @irvinebroque ah! i'm mainly curious about the trade-offs between a client singleton w/ swappable transports vs the "new client per server" approach that MCPClientManager takes - **boristane** (2025-04-27): https://x.com/boristane/status/1916533341911765439 - I just published a new blog post contextual rag on @CloudflareDev using workers ai, vectorized, and d1 it explains everything you need from hybrid search to rerankers and reciprocal rank fusion, with clear code implementation you can copy / paste this and build a saas around https://t.co/1eT9GiKwTD - **LambrosPetrou** (2025-04-27): https://x.com/LambrosPetrou/status/1916526728316596347 - New article out: "Cloudflare Durable Objects are Virtual Objects"🚀 This pops up a lot and is a mindset shift from the traditional way of thinking on "creating" and "destroying" resources, so I wrote a few thoughts how to think about it. https://t.co/dAZE70WZzB #DurableObjects - **BraydenWilmoth** (2025-04-27): https://x.com/BraydenWilmoth/status/1916481531859669436 - Do you have a cool Durable Object project? @janwilmake is making a list of all the cool public projects out there... tag yours here 👇 https://t.co/01vTeaIlcA - **jokull** (2025-04-27): https://x.com/jokull/status/1916456418896384451 - Blog post: connecting AutoRAG from @CloudflareDev to useChat hook from @aisdk https://t.co/Lj55J0v1sS - **mkczarkowski** (2025-04-27): https://x.com/mkczarkowski/status/1916393184088756401 - With @Cloudflare workers and AI-assisted development you can build and deploy an useful MCP server in 15 minutes 🤯 Just built the first version of 10xRules MCP that will let you use the our "Rules for AI" library without leaving your editor. - **pbteja1998** (2025-04-27): https://x.com/pbteja1998/status/1916351497333576158 - https://t.co/NrSEJwMXs7 is now both MCP server creator and MCP server tester (client). You can now immediately test the generated MCP server from within MCPify itself. Using @CloudflareDev agents-sdk to do this automatically. https://t.co/QwP0q2SH7N - **jeff_weinstein** (2025-04-26): https://x.com/jeff_weinstein/status/1916263810173964777 - I want “Save to memory” buttons all over the web. (And a share sheet item to send to memory.) - **ThomAub** (2025-04-26): https://x.com/ThomAub/status/1916238713040080986 - Just listened to @threepointone on @ScalingDevTools and I want to build an AI Agent on @Cloudflare now ! - **theo** (2025-04-26): https://x.com/theo/status/1916212897770049673 - Eric competes in Iron Man challenges, builds best in class dev tools (StackBlitz), and disrupted the status quo of AI app builders (Bolt). He’s also one of the nicest people I know. We need more founders like Eric. https://t.co/JTqVavobuB - **pulsemcp** (2025-04-26): https://x.com/pulsemcp/status/1916190104588325183 - @stuffyokodraws @a16z @apify @rootlyhq @intercom @github (14/14) Read about all this and more: https://t.co/61d1vzxbdF - **pulsemcp** (2025-04-26): https://x.com/pulsemcp/status/1916186366289055991 - (12/14) Who needs a new agent-to-agent communication protocol when @threepointone and @CloudflareDev are out there helping your MCP servers talk to other MCP servers? Love this exhibition of MCP's recursion https://t.co/1a5B0m70ZD - **pulsemcp** (2025-04-26): https://x.com/pulsemcp/status/1916186341836304526 - (1/14) Another week of MCP news in the books. Celebrating the 20th edition of our in-depth MCP newsletter! Highlights: -📊 MCP ecosystem growing 33%+ this month -🪨 More reasons to push MCP Resources - 💅 Turn any website into an editable Figma - 🔎 Pull live https://t.co/jdJMhKCx5Y data in via @ExaAILabs's MCP server More in🧵 - **kentcdodds** (2025-04-26): https://x.com/kentcdodds/status/1916185403876598183 - AI Clients are the new browser MCP Servers are the new websites https://t.co/jasBKypT5J - **chudson** (2025-04-26): https://x.com/chudson/status/1916159959387308167 - Three related trends I’ve noticed lately. The combo of AI-generated cold outreach tools and big lists of VC emails is pushing investors back toward human-driven warm intros and referrals. Partners leaving big firms to start new firms are creating new networks to invest with people they like but maybe couldn’t work with at their old firms based on fund size and strategy. And the density of AI is pulling talent back to SF and consumer startups are consolidating in NYC. In many ways, VC today feels like we are moving back to the way the industry worked prior to 2018. - **liadyosef** (2025-04-26): https://x.com/liadyosef/status/1916155409171206188 - GitMCP now has a built-in online chat! 🚀 You can now chat with the docs online - with the relevant GitMCP server already *pre-installed*! This makes GitMCP the only free, fully open-source documentation tool that has both an MCP server and an online chat. Check it out: - **janwilmake** (2025-04-26): https://x.com/janwilmake/status/1916110534287446079 - Improved performance today Ingesting the entire zip of bun.js (800mb decompressed) and transforming it into a FormData takes 35seconds now. no cache, no filter. On a cloudflare worker with just 128MB of RAM. With filters, it's even way faster! Only get all the markdown from the entire bun repo? 11 seconds. voila! Still have to double check but this looks incredibly promising! - **hichaelmart** (2025-04-26): https://x.com/hichaelmart/status/1915978146282934661 - Published v2.1.0 of fetch-to-node¹. This brings the codebase up to date w/ latest Node.js, chunked encoding support, and streams w/ content-length headers. Perfect if you want to migrate from Express or any other Node.js library, eg for MCP servers ¹https://t.co/DOKIkWjjel https://t.co/PXUgtczNf7 - **pbteja1998** (2025-04-26): https://x.com/pbteja1998/status/1915947288792572179 - https://t.co/NrSEJwMXs7 now supports MCP's new Streamable HTTP transport. Because I am using @CloudflareDev, this is all I had to change to make it work 😅 https://t.co/Rs7ZJBlm9U - **glenmaddern** (2025-04-25): https://x.com/glenmaddern/status/1915905745146745209 - 👀 https://t.co/KfSoXp3yR5 - **G4brym** (2025-04-25): https://x.com/G4brym/status/1915889702218928199 - What if I told you, Django now fully runs on Cloudflare Workers with D1 as the database Will provide more details soon 😃 https://t.co/B043luFiVo - **ankrgyl** (2025-04-25): https://x.com/ankrgyl/status/1915889518118572506 - how many opentelemetry packages does it take to.... - **eastdakota** (2025-04-25): https://x.com/eastdakota/status/1915886840495935848 - I have a BA in English Literature, a minor in Computer Science, a JD, and an MBA. As CEO of @Cloudflare, a fairly technical company, the most important has always been that English Literature degree. To be able to write and communicate effectively, easily, and quickly is the job. https://t.co/23DWKkuGf3 - **cjc** (2025-04-25): https://x.com/cjc/status/1915861381347488076 - @zeeg Explaining all the MCP use cases https://t.co/BXFqOF37Hh - **Baconbrix** (2025-04-25): https://x.com/Baconbrix/status/1915857584722219097 - React Compiler is insane 🤯 Just enabled it in my @Expo Router app and had my mind blown by how much could be optimized when toggling switching screens. Before / After https://t.co/V8MG95VhYD - **elithrar** (2025-04-25): https://x.com/elithrar/status/1915850185873256674 - what else do you even want from a quote?! https://t.co/G4Id4FWWc3 - **zeeg** (2025-04-25): https://x.com/zeeg/status/1915845059519648016 - Got asked for a quote for some press. "MCP is pretty sweet. Cloudflare's support of MCP is pretty sweet. Sentry is pretty sweet. So we made an MCP for Sentry on top of Cloudflare." Don't think marketing is going for it... - **ritakozlov** (2025-04-25): https://x.com/ritakozlov/status/1915820351176544737 - when i say i can't believe how far @cloudflare workers has come along... this is all there was when we started https://t.co/oZ6g0QwI50 - **CloudflareDev** (2025-04-25): https://x.com/CloudflareDev/status/1915819071725072427 - Query & work on data in R2 from the browser thanks to @marimo_io! https://t.co/pDcjE1pf9D - **marimo_io** (2025-04-25): https://x.com/marimo_io/status/1915817443588772338 - ☁️🚀 Deploy marimo Python notebooks to @CloudflareDev as interactive static HTML, in just two lines. Docs: https://t.co/vdAHCbneW9 https://t.co/x0FvZzUC3H - **ritakozlov** (2025-04-25): https://x.com/ritakozlov/status/1915814459668320528 - 💯 https://t.co/auyB8pPJ6I - **threepointone** (2025-04-25): https://x.com/threepointone/status/1915811353949835472 - @alanaagoyal !!! you're hired https://t.co/nWCMbPfXac - **alanaagoyal** (2025-04-25): https://x.com/alanaagoyal/status/1915810920573382711 - out: personal websites in: personal mcp servers https://t.co/gl0DEYkX0N - **fireblaze_15** (2025-04-25): https://x.com/fireblaze_15/status/1915809778774757639 - when i first started using @Cloudflare, it looked like this. crazy that i work here now - nostalgic. https://t.co/Q17ybGhG5b - **dinasaur_404** (2025-04-25): https://x.com/dinasaur_404/status/1915804050144059825 - Great demo showcasing how you can use Durable Objects, Workflows, & MCP to power agent task orchestration! 🤩 https://t.co/ufYBRDShzd - **threepointone** (2025-04-25): https://x.com/threepointone/status/1915803395262206125 - wait WHAT https://t.co/uQ97D3Magy https://t.co/Tlvy1wSfVo - **CloudflareDev** (2025-04-25): https://x.com/CloudflareDev/status/1915800740146122994 - You can actually just scale https://t.co/bhd6ekXcSi - **elithrar** (2025-04-25): https://x.com/elithrar/status/1915794641384853596 - @yagiznizipli tbh never change :) - **ritakozlov** (2025-04-25): https://x.com/ritakozlov/status/1915793729576726533 - you can build MCP servers in python on @cloudflare workers with FastMCP (example below!) which means... you can access durable objects in python now too 😮 - **elithrar** (2025-04-25): https://x.com/elithrar/status/1915793292895170915 - Never skip a weekly Workers demo meeting. https://t.co/pb5TQJJQLA - **kentcdodds** (2025-04-25): https://x.com/kentcdodds/status/1915791554859483411 - Really great example here https://t.co/is8NWGgdBL - **dinasaur_404** (2025-04-25): https://x.com/dinasaur_404/status/1915787019491438653 - Update: This example is now compatible with the ⚡️streamable HTTP⚡️ transport method and has backwards compatibility with SSE. Still just ✨one-click✨ to get the MCP server deployed Shoutout to @JeremyMorrell for handling all the transport complexities so you don't have to! https://t.co/T0AeoaBJSy - **janwilmake** (2025-04-25): https://x.com/janwilmake/status/1915774899244667122 - UIT is OSS, MIT Licensed https://t.co/EEarCdjelf - **janwilmake** (2025-04-25): https://x.com/janwilmake/status/1915774892500300281 - Announcing UIT - performant, modular, low-memory file-processing at scale, in the cloud. It works by offering a 4-step process to gather a file hierarchy from any desired modality, apply filters and transformations, and output it in any desired modality. FULLY OSS (but there's a catch) Check the explanation and video or view it on GitHub 👇🧵 - **ritakozlov** (2025-04-25): https://x.com/ritakozlov/status/1915744299104284952 - mcp2mcp https://t.co/EbcNzgwOWl - **lessin** (2025-04-25): https://x.com/lessin/status/1915738835780067740 - ‘Coding’ is more and more like writing & less and less like ‘building’... App building doesn't make you an engineer, it makes you an author 'Hackathons' are really just write-a-thons. An 'app' is just a business plan / deck by another name. tech 'products' are all consumables. - **threepointone** (2025-04-25): https://x.com/threepointone/status/1915735083278880950 - quick ship: agents@0.0.66 - Agent now has an mcp client by default (this.mcp) that lets you add multiple servers (docs incoming) - (this means, that your mcp server can now connect to other mcp servers. do with that as you will.) - this client now has a helper to generate tools that can be used by @aisdk directly - .stop on useAgentChat now correctly stops the streaming/cancels the request (thanks @lastgoodhandle!) - **yusukebe** (2025-04-25): https://x.com/yusukebe/status/1915674725046792628 - Hono supports Service Worker already. You can run your Hono app on your browser. https://t.co/LEoav33255 https://t.co/OaXVqS1tqo - **theo** (2025-04-25): https://x.com/theo/status/1915641049479024797 - If Anthropic offered business credit cards with a 2% discount on API invoices, I'd sign up for that shit so fast - **brezina** (2025-04-25): https://x.com/brezina/status/1915611776445714563 - @tjparker @jeffawilke Please tell me it’s narrow. 🙏 Like kei truck narrow American vehicles have gotten so damn wide no modern models fit in my narrow urban garage Bring back small cars in America! - **felixnorden** (2025-04-25): https://x.com/felixnorden/status/1915564413853094292 - Builder update on @edenlayer! Task execution and composition through @AnthropicAI's MCP, powered by @CloudflareDev DOs and Workflows for orchestration and long-running MCP workloads - MAGICAL. We're close to full agentic collaboration and are now testing agentic task planning. https://t.co/6gGUQPMnmX - **ritakozlov** (2025-04-24): https://x.com/ritakozlov/status/1915555787100872991 - durable objects + containers is going to be big https://t.co/9kLRmsfohF - **kentcdodds** (2025-04-24): https://x.com/kentcdodds/status/1915545614760173613 - I've got a killer demo to show tomorrow 🔥 - **dok2001** (2025-04-24): https://x.com/dok2001/status/1915520239581724717 - Thrilled to have Tyson join team to help us continue our journey of moving from stateless to stateful! https://t.co/DhJIyR5Yb4 - **dinasaur_404** (2025-04-24): https://x.com/dinasaur_404/status/1915499672963096877 - @kaushikm_ sure will! it's going to be really simple to upgrade your MCP server so that it's compatible with both methods of transport (SSE + streamable HTTP) you'll want compatibility with both since clients have yet to support the new transport method - **kaushikm_** (2025-04-24): https://x.com/kaushikm_/status/1915498557043605604 - @dinasaur_404 pretty neat - will cloudflare go streamable http by default for mcp tools soon? - **ritakozlov** (2025-04-24): https://x.com/ritakozlov/status/1915496927501717610 - watch the cloudflare mcp server repo to see how @cloudflare is building its mcp server(s!!!) in real time mcp as api wrapper is just the beginning of the journey, but coming up with user stories that make it useful is a whole other thing https://t.co/4HP2eiTObj - **CloudflareDev** (2025-04-24): https://x.com/CloudflareDev/status/1915482827270902089 - Deploy (your own MCP server) to Cloudflare 😎 https://t.co/KqsH9VcIob - **dinasaur_404** (2025-04-24): https://x.com/dinasaur_404/status/1915479655269753159 - Just ✨one-click✨ to deploy a remote MCP server on Cloudflare! We posted an example of an authless MCP server that you can start using right away. Customize your tools, test it out in the AI Playground, and add auth later (if/when needed). https://t.co/wJQMWa53eN - **thebatdev** (2025-04-24): https://x.com/thebatdev/status/1915477140310831141 - thinking of publishing some @SST_dev templates that I've been using for quick 0-1 for my projects. I think a lot of devs should get to know how easy and efficient SST is, I don't manage any env files now, i hop from my local machine to my server with ease. I can share my repo with others and they can get setup in minutes with no fuss. I can even manage a mobile app built using expo and RN with Dev commands again, no env required. all that in a monorepo. and even after that if i don't want to deploy a backend and just want to use secrets i can just do that. - **kristianfreeman** (2025-04-24): https://x.com/kristianfreeman/status/1915455394786533582 - This is an awesome project built with Workers and Hono, worth a watch if you build with @shadcn UI! https://t.co/7DDekixWGG - **developedbyed** (2025-04-24): https://x.com/developedbyed/status/1915444376332902435 - super fun time working with @honojs and @Cloudflare workers to serve your own shadcn components 💖 https://t.co/7XgLe1vOik https://t.co/oYWSOOLRdu - **ScalingDevTools** (2025-04-24): https://x.com/ScalingDevTools/status/1915439420544557422 - AI Agents special with @threepointone 00:16 React and AI agents 11:20 Sunil's 9 to 5 theory of automation 17:05 It's inevitable that AI agents take over 22:38 Durable objects 28:48 MCP servers 30:41 Chat is going to dominate as the UI for agents 32:55 Granola and collaboration with AI agents 35:15 What should people be building 38:09 DevTools opportunities: AI native devtools 42:52 Cloudflare, PartyKit, Cloudflare acquisitions and team - **threepointone** (2025-04-24): https://x.com/threepointone/status/1915433832091693206 - I'm going to talk about roadmap, and maybe a small demo of something I just did (hint: MCP!) https://t.co/1w1WzQzpqE - **RedwoodJS** (2025-04-24): https://x.com/RedwoodJS/status/1915417425371185335 - Most frameworks treat the document as a global: a html page with a head, body, etc, a one-size-fits-all concern. RedwoodSDK treats it as just another component - one you control. Want a different document per route? You're in control! https://t.co/UAc9IIuIH4 - **TysonTrautmann** (2025-04-24): https://x.com/TysonTrautmann/status/1915415364026150964 - Excited to share I've joined @Cloudflare as Senior Director of Engineering leading storage & database products like R2, Hyperdrive, Vectorize & KV! Love the mission to help build a better internet and I'm stoked to be a part of it. 🚀 - **jplhomer** (2025-04-24): https://x.com/jplhomer/status/1915402278028169329 - Stuff I'm going to play with as soon as I find time: - @samgoodwin89's Alchemy (kinda like a TS-native SST) - @RedwoodJS SDK (aka my dream of a fullstack toolkit built on CF) with RSC support Cool stuff happening here. Enjoyed chatting a bunch with Sam at React Miami! https://t.co/o46nEZZ8OX - **ibuildthecloud** (2025-04-24): https://x.com/ibuildthecloud/status/1915396856433303789 - @luis11181a For most simple things yes, but bidirectional communication unleashes the full power. I already know because I've been realistically doing this crap for some time and you need it for certain things. Sampling is the biggest one. But roots is equally important. - **jplhomer** (2025-04-24): https://x.com/jplhomer/status/1915386710269087860 - One of my fav pieces of React Miami swag tbh https://t.co/9dpofxyLVB - **kentcdodds** (2025-04-24): https://x.com/kentcdodds/status/1915337998180221017 - @ibuildthecloud I've cracked this issue with a itty-bitty hack that seems to be required if you want to do registration and authentication 100% within the MCP flow. https://t.co/9CfVW91zB8 https://t.co/fCi7Y20xQM - **michaelxpettis** (2025-04-24): https://x.com/michaelxpettis/status/1915271417295229013 - 8/8 stave off overcapacity is temporarily, with rising debt. Once debt is no longer able to rise, as we saw in the late 1920s/1930s (and as Marriner Eccles explained), overcapacity must be curbed, which, usually means a drop in production and a contraction in global trade. - **michaelxpettis** (2025-04-24): https://x.com/michaelxpettis/status/1915271414375924152 - 7/8 China (and everyone else) should invest as much as possible in productivity-enhancing technology because only increases in productivity lead to sustainable increases in wealth, but if increases in overall productivity are not matched with wage increases, the only way to... - **samgoodwin89** (2025-04-24): https://x.com/samgoodwin89/status/1915257205156368478 - Neon Project + Cloudflare Hyperdrive + TanStackStart in 10 lines of code. https://t.co/SKRrhIxobM - **t_blom** (2025-04-24): https://x.com/t_blom/status/1915226122859798916 - I’ve noticed that when founders fail at sales & marketing, it’s often because they’re cosplaying as a (bad) salesperson. Instead of asking themselves “what kind of message would I be interested to receive”, they flip into “sales mode” and spout garbage. “Our product is a paradigm shift that will synergize perfectly with your internal and external stakeholders” 🤮 - **RichardHanania** (2025-04-24): https://x.com/RichardHanania/status/1915196256512888959 - I’ve never seen a policy debate like this. The market is telling you in real time, obviously and consistently, what is good policy and what is bad policy. If you support these tariffs you’re simply unreachable via empirical data. https://t.co/LlpKIWFt2M - **jonas** (2025-04-23): https://x.com/jonas/status/1915126431538975026 - Good examples of vitest test harnesses for workers with D1. Want to set up isolated seed data for each of my tests (or all tests in a group) and then call functions on the worker Extra bonus points if it also shows a golden path for mocking other workers that I have bindings to (but unrelated to D1) - **MaximeServais77** (2025-04-23): https://x.com/MaximeServais77/status/1915125418740396465 - @burcs @CloudflareDev I wasted a bunch of time because of a simple config issue - would really add a warning to the docs: https://t.co/IpBw7xWs4I - **burcs** (2025-04-23): https://x.com/burcs/status/1915124469708079463 - currently adding some getting started docs inside the Cloudflare dashboard what other guides or examples would you find helpful? https://t.co/DEeK5453cB - **jonbeckman** (2025-04-23): https://x.com/jonbeckman/status/1915064983966072903 - I have big plans to democratize monitoring your website, your APIs, your whatever. Unapologetically 100% Cloudflare because it's the best place to do this. Not a VPS. Not AWS. https://t.co/hLwckol6t1 - **dinasaur_404** (2025-04-23): https://x.com/dinasaur_404/status/1914843005313307064 - agents of chaos 😛 @ritakozlov @elithrar https://t.co/bRJkNJqXUA - **bertie_ai** (2025-04-23): https://x.com/bertie_ai/status/1914832347343822915 - Already cooked at @wordware 👨‍🍳 Using @Cloudflare Durable Objects https://t.co/dqi7UCJ1v7 - **elithrar** (2025-04-22): https://x.com/elithrar/status/1914791626775445766 - Picture it: it's Monday night, June 2nd. You're in San Francisco. You have this strange urge to chat about Big Data, streaming ingestion + modern data formats. Well, turns out we have exactly the thing: come by Cloudflare HQ for a Big Data meet-up! https://t.co/9TZpcS1d7r - **threepointone** (2025-04-22): https://x.com/threepointone/status/1914695108231860395 - @cramforce landing in agents as well! https://t.co/1jpPFTWFpt - **Bfaviero** (2025-04-22): https://x.com/Bfaviero/status/1914507677444477105 - Crazy how NYC in 1931 looks like SF today https://t.co/vdvzWBZYD2 https://t.co/k4rrugBUgG - **Jesse_Livermore** (2025-04-22): https://x.com/Jesse_Livermore/status/1914506441269932157 - If you think people are angry about the Trump 2.0 shitshow now, just wait until average middle-class voters find out that they're going to have to pay higher prices for everything at the store during a self-induced recession that might cost them their jobs. - **appfactory** (2025-04-22): https://x.com/appfactory/status/1914480521729777928 - For a long time, software was something you used made by companies, served to you. Then the tools got better. The platforms got cheaper. AI lowered the barrier. And suddenly, people didn’t need permission. They started writing their own tools again: Not just apps, but custom scripts, internal systems, weird ideas made real. We call these people 'high-agency.' But I call what they make something else: Personal Software. It’s not built to scale. It’s built to serve. It doesn’t raise a Series A. It scratches an itch. It’s not SaaS. It’s yours. - **lessin** (2025-04-21): https://x.com/lessin/status/1914450043425427506 - The state of VC in 2025. Where we are, where we are going, and where to invest ... https://t.co/NvOOYzeEpG https://t.co/cbXha8w19L - **theo** (2025-04-21): https://x.com/theo/status/1914427848380834267 - Code review is an unsolved problem - **matthewcp** (2025-04-21): https://x.com/matthewcp/status/1914380539202846758 - There's not enough people experimenting with Actor model on the frontend. If there's anything that will unseat today's reactive programming reign, it will probably include actors. - **ryanflorence** (2025-04-21): https://x.com/ryanflorence/status/1914368968518983900 - @ebey_jacob Almost all of the code to enable "Route Module API" like server loader, client loader, middleware, code splitting, links, meta, etc. is in the Vite plugin With RSC it's almost all in our runtimes "Data Mode" stuff or can simply go away Pretty slick - **ryanflorence** (2025-04-21): https://x.com/ryanflorence/status/1914368015040385125 - Oh, I missed one in the server client handoff: - data - client assets - action payloads When we release this stuff, I hope to show how simple @ebey_jacob's work in React Router makes it to just deploy React however you want - **ryanflorence** (2025-04-21): https://x.com/ryanflorence/status/1914366747794399418 - RSC enables React Router v7 Framework to be even less dependent on our first-party Vite plugin, opening up much simpler integration with other bundlers. Now React prescribes the server → client handoff (data, client assets) so we can move vite plugin logic into our runtimes - **Esteban_Puerta9** (2025-04-21): https://x.com/Esteban_Puerta9/status/1914351935575990745 - Cloudflare has remote mcp auth (no one else is doing this yet) AI gateway and a whole AI offering and now a container platform?! people need to start looking at Cloudflare as the de-facto dev platform to build on. 👋 bye AWS https://t.co/JTGtrGqNUY - **BraydenWilmoth** (2025-04-21): https://x.com/BraydenWilmoth/status/1914327561837949159 - Want to make your D1 API requests up to 6x faster? Deploy this worker to get CRUD REST table API plus the ability to send raw queries.. and it's blazing fast. > Execute queries > REST table API > Sort, order, limit, offset+ > Secret Store for Authorization Github in 🧵 https://t.co/kYwUTknn36 - **elithrar** (2025-04-21): https://x.com/elithrar/status/1914320646936371681 - a full-stack demo app for @cloudflare Workers + React + @PlanetScale would be pretty cool, right? https://t.co/8RxQTrXcJB - **kentcdodds** (2025-04-21): https://x.com/kentcdodds/status/1914319321494380726 - I'm building an app that is 100% only accessible via your AI Assistant using MCP. My server sends no HTML at all. Including authentication and payment. AMA - **ankrgyl** (2025-04-21): https://x.com/ankrgyl/status/1914161948910121220 - the fact that k8s still exists and thrives is the exception, not the norm. shitty software does not usually make it. - **cremieuxrecueil** (2025-04-20): https://x.com/cremieuxrecueil/status/1914103137570275698 - Really neat paper! It says a lot: H-1Bs make America richer, Argentina's capital controls cost it billions, quotas on Chinese textiles cost the U.S. billions, approving more taxi medallions would make NYC richer, and most of all: San Francisco has the most to gain from upzoning https://t.co/Ys20zNpJ6V https://t.co/dA6PpiqMHI - **JonasBadalic** (2025-04-20): https://x.com/JonasBadalic/status/1914096796801880533 - I cant stand the shit that kids are listening to these days man… I’m playing ball on a warm day in NYC, 5min away from the spots that produced the most iconic hip hop, and this is the best y’all can do? - **boristane** (2025-04-20): https://x.com/boristane/status/1913784863259230276 - omg https://t.co/OqbQmCoNa5 - **ankrgyl** (2025-04-20): https://x.com/ankrgyl/status/1913776821537329270 - RAG is one flavor of context engineering. For those that say “RAG is dead” hopefully this framing makes it obvious why that’s not the case. - **warriorsworld** (2025-04-20): https://x.com/warriorsworld/status/1913773425031868706 - https://t.co/U61Y33XTaP - **fireblaze_15** (2025-04-20): https://x.com/fireblaze_15/status/1913769958569296239 - shipped few weeks back ~ you should now see account level metrics in your @Cloudflare R2 dashboard ✨ let me know your top requests for R2 dashboard! 👇👇 https://t.co/XjLJLFSdy2 - **ankrgyl** (2025-04-20): https://x.com/ankrgyl/status/1913766591910842619 - As models get more powerful, i find myself focusing more effort on context engineering, which is the task of bringing the right information (in the right format) to the LLM. Context engineering is hard because it is pervasive. You need to engineer every layer of the stack to capture and make context available. Send too little context, and the LLM won’t know what to do. Too much, and you’re out of tokens or the LLM gets lost. Good context engineering caches well. Bad context engineering is both slow and expensive. - **dok2001** (2025-04-20): https://x.com/dok2001/status/1913758745911341323 - @burcs Ship it - **burcs** (2025-04-20): https://x.com/burcs/status/1913756726471320024 - i dream of a simpler web... https://t.co/tR1eYeF5Cr - **paulg** (2025-04-19): https://x.com/paulg/status/1913736632408346671 - @cosmicfibretion @MabreyTed Yeah, I know. It would be so easy. One sentence that any genuinely patriotic US citizen would have no trouble saying. - **paulg** (2025-04-19): https://x.com/paulg/status/1913729703447241087 - @MabreyTed As I said, I'll be happy to delete it if you commit publicly on behalf of Palantir not to build things that help the government violate the US constitution. Will you do that, Ted? - **jonbeckman** (2025-04-19): https://x.com/jonbeckman/status/1913728922597859826 - v1 is here. Get it up and running in less than 10 minutes. More updates coming soon. https://t.co/mMl4VZ7xRQ https://t.co/rPd0ObCK1Q - **zebassembly** (2025-04-19): https://x.com/zebassembly/status/1913707981625262316 - This one is a little complicated since the algorithm being (relatively) slow is actually by design and we limit CPU time on the free tier. Looking through better auth and its deps, it uses a mix of runtime-provided crypto and JS implemented crypto with I suspect most of it being the latter. With it being implemented in JS, depending on how long your Worker has been in memory the JIT compiler might or might not have optimized your code (why some people are seeing 5s vs ~200ms for signups). Crypto for accounts being slow is usually a feature to prevent brute force attacks if your DB ever gets compromised, but because this is JS it’s extra slow compared to the optimized code a would-be attacker would actually use. Better-auth could use native crypto (node:crypto via nodejs_compat maybe?) for Workers but even then it probably wouldn’t be faster than the 10ms of CPU limit on the free tier. - **aninibread** (2025-04-19): https://x.com/aninibread/status/1913679585289322618 - Haha you beat me to it!! So excited to see filters land in AutoRAG 🗂️ — huge thanks to @G4brym for the early drop. Makes it way more powerful for multi-tenant use cases! https://t.co/gR8gk0aOX3 - **TysonTrautmann** (2025-04-19): https://x.com/TysonTrautmann/status/1913644707235958878 - An important insight from this thread is that MCP isn't just standardizing API formats - it's designing interfaces that work within today's LLMs' cognitive constraints. The challenge is creating APIs that models can actually reason about an use correctly and repeatably. https://t.co/DQY80Gk0f5 - **pulsemcp** (2025-04-19): https://x.com/pulsemcp/status/1913613518043222067 - (5/10) There's an open PR from Anthropic to add an outputSchema to Tool calls. This option to predefine the structure should help unlock opportunities for MCP client apps and agents to do things with Tool calls beyond just plug them into conversation context. https://t.co/ckInPyAnMU - **pulsemcp** (2025-04-19): https://x.com/pulsemcp/status/1913613511739294027 - (2/10) We like @kentcdodds's idea to kick off a discussion on bringing more UI to MCP. Imagine your MCP client app apps displaying structured data around tool calls, like in Claude's artifacts, but with no latency and potentially interactivity too. https://t.co/haWkWTarP1 - **pulsemcp** (2025-04-19): https://x.com/pulsemcp/status/1913613509403033785 - (1/10) This week on the bleeding edge of MCP news, read about: - 📈UI in MCP - ⚙️MCP server configuration schema - 🟩Return shapes for tool calls - 🛠️Grouping tools as "interfaces" And more 🧵 https://t.co/vQfrnWcc0P - **liadyosef** (2025-04-19): https://x.com/liadyosef/status/1913611902086696994 - Ok, that's amazing. 🦄 A side by side three.js comparison - with GitMCP and without. GitMCP's story originally started from a tweet by @threejs about their large, non-LLM-digestable, documentation file. @idosal1 and I ended up creating a cool generic documentation MCP server, but we didn't forget threejs and @mrdoob. So we put it to the test. Using GitMCP, I gave this prompt to Cursor + Claude 3.7: "Build a Three.js scene featuring a controllable realistic person navigating a textured dynamic urban environment with realistic lighting and subtle bloom effects. Ensure keyboard controls (WASD) for movement." This is the result. To the right is the one-shot result with our MCP server. To the left is the result without it, based just on Cursor's limited knowledge. The video is pretty conclusive. 🤯 It's really amazing since Cursor already knows a lot about three.js, but with our dedicated documentation MCP server it just outputs better results. And for all other libraries out there - that are not included in Cursor - this is a total game changer. A huge shout out to @CloudflareDev and @aninibread, who worked with us yesterday to quickly insert new capabilities in their AutoRAG feature (that enables indexing large amount of documentation) according to our specific needs, and to troubleshoot issues. It was very helpful. Knowledge is power, and that applies for every coding assitant. And not just for three.js! Check out GitMCP for any library you're using - link in the next comment. - **ghostwriternr** (2025-04-19): https://x.com/ghostwriternr/status/1913525653531435301 - Coding agents are easy. This week, I prototyped one using only @aisdk, @CloudflareDev's agents SDK and their upcoming containers platform (to access filesystem + terminal). And giving it only 6 tools. It's already able to build itself! Remote agent fleet for all of us? 🔗 https://t.co/zpHGGCHNzc https://t.co/VR5UocMyrb - **ghostwriternr** (2025-04-19): https://x.com/ghostwriternr/status/1913525655758655821 - https://t.co/hanKOiAX1U - **backpinelabs** (2025-04-19): https://x.com/backpinelabs/status/1913434393760190718 - If anyone is wondering how to ship a @tan_stack start application to Cloudflare Workers (with access to CF bindings) I have a working example. Bindings are accessible in production and locally even when using vinxi dev. https://t.co/MlCw2p5BYA - **markdalgleish** (2025-04-19): https://x.com/markdalgleish/status/1913394010389159937 - Same here. Back when I was getting started, @ryanflorence was one of the few people who took JavaScript seriously and blogged about it. When I was getting into @reactjs and found he was already building a router for it, I was so excited. And now I get to work on it with him. https://t.co/WqvaLAZSxn - **yagiznizipli** (2025-04-19): https://x.com/yagiznizipli/status/1913393568770650386 - At this point, I know more about tcp connections than my knowledge on Swedish language. That’s just sad… - **yagiznizipli** (2025-04-18): https://x.com/yagiznizipli/status/1913379025826820480 - In case there is a person who wants to connect mssql from @Cloudflare workers, soon you can! - **ankrgyl** (2025-04-18): https://x.com/ankrgyl/status/1913356044807663728 - The #1 most useful thing I learned early on as a programmer: When something doesn't work, assume I am wrong. Not the computer. Not the library. Not my colleagues. Not the vendor. Me. - **G4brym** (2025-04-18): https://x.com/G4brym/status/1913352119840547220 - Sorry @aninibread, but I couldn't wait for Monday to publish that AutoRAG now supports filters 🚀🎉 *new rags* can now filter results by folder and timestamp Public docs and support for existing rags will be live next week https://t.co/TWijVoshmM - **paulg** (2025-04-18): https://x.com/paulg/status/1913338841068404903 - It's a very exciting time in tech right now. If you're a first-rate programmer, there are a huge number of other places you can go work rather than at the company building the infrastructure of the police state. https://t.co/Wf33KHag79 - **zeeg** (2025-04-18): https://x.com/zeeg/status/1913301449498312739 - @0xferruccio 100% - we have the same problem with our APIs. Some places we only support project IDs (numeric) and its awful. You really gotta optimize them fit closer to a model of how you'd reason about a UI vs a machine API. - **0xferruccio** (2025-04-18): https://x.com/0xferruccio/status/1913300713125433364 - @zeeg Yeah for example the Slack MCPs I've used all use uuids for CRUD operations like posting messages to a channel because of how the Slack API works But it's very unlikely that if one was designing APIs for easy consumption by LLMs that we'd have the same required inputs - **jonbeckman** (2025-04-18): https://x.com/jonbeckman/status/1913292945609867776 - This is how it works, the trigger (Durable Object + Alarm API) and the executioner (Worker). Each endpoint gets a DO that acts as a programmable cronjob. waitUntil() allows the Worker call to return immediately so the DO is not waiting. The Worker makes the health check async. https://t.co/gT1McpLiBx https://t.co/rPd0ObCK1Q - **zeeg** (2025-04-18): https://x.com/zeeg/status/1913289984875901133 - This gets even worse when you provide a lot of tools - say for example, every API endpoint available in your OpenAPI spec. In my test with just a dozen tools, two of which are _slightly_ similar, it routinely failed to call the right tool unless you used precise language. - **zeeg** (2025-04-18): https://x.com/zeeg/status/1913289981520490558 - For example, say you have a "create_post" endpoint. It probably has a ton of optional parameters to it. You might have rough explanations for each parameter. Which ones is the LLM going to fill in? Are you going to change your API to optimize it to LLM guidance? Its not enough. - **zeeg** (2025-04-18): https://x.com/zeeg/status/1913289982665519398 - What I mean by that, as I discovered yesterday, is LLMs are constraining context consumpion from tools (for good reason). For example, I believe 4o limits your tool description to 1024 characters. Thats not even enough to guide LLM-specific, super constrained APIs. - **zeeg** (2025-04-18): https://x.com/zeeg/status/1913289979368796671 - A big mistake people are making is thinking MCP is just a shim on top of your existing API. That is, people are proxying MCP <> OpenAPI - no diff than codegening it. This is absolutely broken by design, isnt useful, and is a waste of your time. This is my perspective on why. - **kentcdodds** (2025-04-18): https://x.com/kentcdodds/status/1913289331730702512 - Oh, hey there! I want you to join me for a workshop. Last time this sold out in only 5 hours (‼️) and I've got a 40% discount on this workshop until Sunday, so don't delay! See you on the 15th of May! https://t.co/YyPlyystyC - **mrlincolnlogs** (2025-04-18): https://x.com/mrlincolnlogs/status/1913254155793109101 - Are you afraid of the dark?? Coming soon to a @Cloudflare dashboard near you 😱 https://t.co/AGUp8rRspZ - **ritakozlov** (2025-04-18): https://x.com/ritakozlov/status/1913238026227957854 - aight, back to our regularly scheduled content... a couple weeks ago we release support for playwright on cloudflare browser rendering, and we just added an MCP server to interact with it! link to repo 🔗 https://t.co/4gIXjKQPZw - **jonbeckman** (2025-04-18): https://x.com/jonbeckman/status/1913226819723055569 - Should I open source this? Been building this uptime monitor that's based on @CloudflareDev Workers, which makes it incredibly cheap to run. In fact it's 51x(!) cheaper per check than the popular @checklyHQ https://t.co/xK27EJr64M - **ryanflorence** (2025-04-18): https://x.com/ryanflorence/status/1913039098245419184 - Remix team (@ebey_jacob) figured out how to get RSC into React Router without splitting the runtime the way pages/app router do in Next Works for both "RSC first routes" and "RSC from loaders" - **danielkelly_io** (2025-04-17): https://x.com/danielkelly_io/status/1912916133637746901 - In the age of AI, this is a really smart docs feature. As seen on @Cloudflare https://t.co/4Yxh7lkEVa https://t.co/oxDAgcpoiQ - **thomasgauvin** (2025-04-17): https://x.com/thomasgauvin/status/1912915511156879634 - Little treat post-@CloudflareDev Week: Bulk reads in Workers KV just landed! Now, you can fetch multiple key-value pairs using the Workers KV binding in a single operation 🚀 TLDR: faster and easier to get multiple values in a single Worker invocation! https://t.co/1Tb7BgXFgK - **typesfast** (2025-04-17): https://x.com/typesfast/status/1912890256216695002 - Thousands, and then millions, of American small businesses, including many iconic brands, will go bankrupt this year if the tariff policies on China don’t change. 🧵 - **thedanigrant** (2025-04-17): https://x.com/thedanigrant/status/1912873082198565153 - My 1st day at Cloudflare, the now CTO gave me 1 big piece of advice: Ship small wins early. Like ASAP. The easiest way to build trust in a new org is to ship small wins early. Ship small and fast (otherwise you're just the new PM working on the same darn feature for 6 months). - **elithrar** (2025-04-17): https://x.com/elithrar/status/1912872295011410145 - gotta start them young (learning how to build on Workers) https://t.co/DABbF6cm6d - **dinasaur_404** (2025-04-17): https://x.com/dinasaur_404/status/1912871552770576730 - and very soon we'll have streamable http support for remote mcp servers built on cloudflare 🫢👀 https://t.co/SsmVzYyAle - **glenmaddern** (2025-04-17): https://x.com/glenmaddern/status/1912806730410656046 - Btw `mcp-remote@next` is available for initial testing if you're building MCP servers using the new Streamable HTTP transport. It's pretty rough, but it does seem to gracefully fall back from SSE -> HTTP (or vice versa) https://t.co/AwnYNpJjGd https://t.co/dyXlkMdaW8 - **nwbotha** (2025-04-17): https://x.com/nwbotha/status/1912801391396991340 - I find myself defaulting more and more to Cloudflare when hosting new services and projects. Previously, it was Vercel and DigitalOcean, but Cloudflare’s offering has become so good over the past couple of months that it’s now the first option I consider when building something. - **samlambert** (2025-04-17): https://x.com/samlambert/status/1912708352762884460 - if i need entertainment during the day i go to the blue bottle in south park just to overhear some of the worst takes on planet earth - **paulg** (2025-04-16): https://x.com/paulg/status/1912626115035951297 - "Every hardware startup's working capital needs just doubled." — founder of a hardware startup - **samgoodwin89** (2025-04-16): https://x.com/samgoodwin89/status/1912589644707049658 - You can now deploy TanStack Start to Cloudflare Workers+Assets with Alchemy. https://t.co/zbCdZbxyAO https://t.co/CGO86coR5f - **RedwoodJS** (2025-04-16): https://x.com/RedwoodJS/status/1912517284595105858 - Built on Standards The Web Platform is the framework. Request and Response are the native API. Stream responses, upgrade protocols, debug in DevTools. No abstractions, no black boxes. - **charl_dot_dev** (2025-04-16): https://x.com/charl_dot_dev/status/1912467926231269700 - Consume github repo in Cloudflare R2 using a Cloudflare Agent and whack AutoRag on it, pretty impressive... and simple https://t.co/UPTOdrK8nH - **levelsio** (2025-04-16): https://x.com/levelsio/status/1912460457052434874 - Another bug submitted by a user, that I just copy pasted into Cursor and then immediately fixed I think I need to create an automated workflow that pulls bugs reported from my @featurebase bug board, then sends them into Cursor/AI etc, to immediately fix it, then do a pull request that I can then check Fully automated AI bug fixing workflow will be the standard I think And then my daily work will just be checking the AI's pull requests and approve or reject - **bennpeifert** (2025-04-16): https://x.com/bennpeifert/status/1912341433283145791 - This is not a surprise attack. They've told you exactly what they're planning from the very beginning. Stop being surprised. Stop telling us we're over-reacting. This is how democracies fall. https://t.co/nxfW6xESL3 - **samgoodwin89** (2025-04-16): https://x.com/samgoodwin89/status/1912318184960717204 - Gotta set the record straight on this one, I needed to set a later compatibility date. User error! But hey, now we got pipelines in Alchemy, yeah boiii. https://t.co/rGK0D2AD4l https://t.co/ZZd3MV9vfw - **stytchauth** (2025-04-15): https://x.com/stytchauth/status/1912282785789911440 - Remote MCP servers meet authentication and AI Agents 🚀 Stytch + Cloudflare break it down in this demo— check it out 👇 https://t.co/8uBJPtP9ge - **mustafaakin** (2025-04-15): https://x.com/mustafaakin/status/1912277388961984770 - Working with containers, kubernetes made me so tired all of the years. I just want to write everything on Typescript now and deploy to Cloudflare. - **batuhan** (2025-04-15): https://x.com/batuhan/status/1912273683504046383 - @CloudflareDev multiple folks from cloudflare reached out. they are on the case! thank you @threepointone @dinasaur_404 @irvinebroque tbh this bug made me even more confident to ship on cloudflare, it's really awesome to see how much they care - **NickTimiraos** (2025-04-15): https://x.com/NickTimiraos/status/1912247275222421623 - United Airlines takes the unusual step of providing guidance under two different economic scenarios: One if the economy enters a recession, and one if it doesn't. https://t.co/ofd7153nOS https://t.co/CW0drFDNvU - **elithrar** (2025-04-15): https://x.com/elithrar/status/1912170142462517755 - FYI: I'm hiring for a Sr PM to join our Data Platform team! If you know what Iceberg is (it's not the thing that sunk the Titanic) + know your OLAP from your OLTP + care about making working with structured data easier for app devs, then apply here: https://t.co/5PRktYeCDJ https://t.co/0pgttzEIy3 - **tracyalloway** (2025-04-15): https://x.com/tracyalloway/status/1912146150389399786 - I told @TheStalwart I’d take him to one of the best restaurants in London. A really underrated spot https://t.co/rlXpaB2B98 - **KentonVarda** (2025-04-15): https://x.com/KentonVarda/status/1911960700953506294 - Got my 2025 property tax assessment today and... it's... down? What? Yes, apparently, across the city, assessments are down an average 4.5%. I guess this is because Austin has been building tons and tons of housing. Funny how that works. - **dinasaur_404** (2025-04-14): https://x.com/dinasaur_404/status/1911848856175726643 - another remote MCP server built on Cloudflare Workers 🤩🧡 keep em comin! https://t.co/GeTvSvrfDW - **kentcdodds** (2025-04-14): https://x.com/kentcdodds/status/1911827457268756782 - In about an hour and a half I'm going to be joined by @threepointone to talk about AI Agents, MCPs, and the future of user interaction. It's gonna be a good time! Join us on YouTube! https://t.co/14t7iYWk0X - **liadyosef** (2025-04-14): https://x.com/liadyosef/status/1911808540274045140 - * CISOs around the world suddenly wake up with a cold sweat * https://t.co/xpVujS0Lti - **idosal1** (2025-04-14): https://x.com/idosal1/status/1911800845080879207 - It's been a week since @liady and I released GitMCP, and the response has been amazing! 1. #1 Top Trending on @github with 1.2K stars! 2. Headline front page article for almost two days in the leading tech magazine here (@geektime) 3. The default server in MCP Garden Inspector 4. Adopted by projects as the default way to access their documentation (@ingonyama) 5. Collaborations with some of the largest libraries will soon be announced! 6. Shoutout from @jeremyphoward, @GithubProjects, the leading MCP newsletter @pulsemcp, and more! 7. Amazing feedback from @dinasaur_404 and the @CloudflareDev team We're humbled by the community's feedback and the critical gap that GitMCP fills. This is just the beginning. GitMCP gets better every day, with lots of cool stuff coming soon. Stay tuned! - **BraydenWilmoth** (2025-04-14): https://x.com/BraydenWilmoth/status/1911794824770142603 - Cloudflare D1 users – I want your feedback. Tell me what is missing in the dashboard or user experience today. What is a feature/thing you'd love to see that would help make the offering _world class_? Don't hold back. - **RedwoodJS** (2025-04-14): https://x.com/RedwoodJS/status/1911786485407232386 - Every Route Is Just a Function Return JSX, stream a response, or upgrade to websockets. There's no special syntax or compiler magic. A route is a standard function that takes a Request and returns a Response( or even a ) - **laiso** (2025-04-14): https://x.com/laiso/status/1911780595740578231 - Remote MCPとcloudflare/puppeteer https://t.co/qA0GIK0LYT でBroweser Use Cloudを自作できそうだね - **michellechen** (2025-04-14): https://x.com/michellechen/status/1911761000728244447 - coming off the weekend of dev week and feeling so grateful for the workers ai team. on top of dev week we had a surprise saturday Llama 4 launch, a big customer event, and more - it’s been a long week. giving a big virtual hug to the team @ashishbdatta @davzaman @kevinjain_ @xjewer @svensauleau @celso @ritakozlov + others not on twitter but equally appreciated 🧡 these are the real brains behind the launches - **pbteja1998** (2025-04-14): https://x.com/pbteja1998/status/1911723180445610149 - Finally (re)launching https://t.co/NrSEJwMpCz 🚀 MCPify is like Lovable/Bolt/V0, but for building MCP servers! You just talk to the AI and can create your own MCP server. I built this completely using @CloudflareDev Cloudflare Workers + Durable Objects + Workers for Platforms https://t.co/kPBmIGE6bL - **threepointone** (2025-04-14): https://x.com/threepointone/status/1911713088404889884 - new post: on guitars and javascript a love letter to my 2 favourite technologies. link in reply. https://t.co/Enn6EiYbYu - **batuhan** (2025-04-14): https://x.com/batuhan/status/1911710324832563666 - i wrote a silly little MCP with Cloudflare's libraries and it wasn't fun because it was done under 10 minutes it just a thing that exposes DNS queries + whois over MCP. add to your cursor today!!! it's at ne(.)tools you can ask it what CDN or email provider a domain uses, check if things are setup correctly. search for domains with AI - **invisal89** (2025-04-14): https://x.com/invisal89/status/1911692246182928753 - SQLite Internal — a "very human" SQLite file format viewer — is now available! It's still a work in progress, but I’m releasing it early to gather feedback. https://t.co/2kNFq9yKhr https://t.co/KnDzyISXcy - **boristane** (2025-04-14): https://x.com/boristane/status/1911585853056385305 - just like that I have a datalake I can query fully integrated in my app to capture all user actions in just about 30mins I had never used neither @cloudflaredev pipelines nor duckdb (with @motherduck) 2025 is fantastic https://t.co/SnxXDHWscx https://t.co/GcPDfHlLDj - **boristane** (2025-04-14): https://x.com/boristane/status/1911574344452767874 - time to try out @CloudflareDev pipelines I'll use it for: - an audit log of all user actions, and surface the logs to users on demand - a datalake for analytics https://t.co/BU7Q7XOxvn - **ericzakariasson** (2025-04-13): https://x.com/ericzakariasson/status/1911568902758490577 - one example of how to leverage existing tooling with @cursor_ai agent terminal here we find all the broken links and have agent go through them one by one https://t.co/rchrKdNrkq - **tchayen** (2025-04-13): https://x.com/tchayen/status/1911546279785845210 - If you know about JS bundlers more than you would like to admit and want to help thousands of developers build some of the best App Store apps that reach millions of users – apply below ↓ https://t.co/Rp8OEL8WuM - **Baconbrix** (2025-04-13): https://x.com/Baconbrix/status/1911544140803760317 - I’m hiring for the 𝝠 Expo Router project! We’re building: ◆ Universal file-based routing ◆ One-click native & web deployment ◆ Expo for web ◆ Server-driven UI Come build the universal React framework and push mobile dev to its absolute limits! 🚀 https://t.co/T6kFYLypoZ - **_ashleypeacock** (2025-04-13): https://x.com/_ashleypeacock/status/1911523914649415881 - I've been working on something for the devs who live on the edge (see what I did there?) 🔥 I'll drop the first content piece tomorrow, but if you want to sign up early, the link is below. It's totally free, and there will be no subscription or locked content. https://t.co/eMBFsiFybe - **RedwoodJS** (2025-04-13): https://x.com/RedwoodJS/status/1911489741054525833 - RedwoodJS is a React Framework for Cloudfare. It begins as a Vite plugin that unlocks SSR, React Server Components, Server Functions, and realtime features. Its standards-based router, with support for middleware and interruptors, gives you fine-grained control over every request and response. - **backpinelabs** (2025-04-13): https://x.com/backpinelabs/status/1911477531947254197 - Cloudflare Workers with asset routing makes building and deploying React applications stupid simple. This videos showcases how to build upon @CloudflareDev React on Workers template with @tan_stack, @trpcio, and @shadcn https://t.co/RYR5HPDPqD - **conorsen** (2025-04-13): https://x.com/conorsen/status/1911461759430656367 - You see the point made every so often but “the cost of Volcker crushing inflation in the 1980’s was hollowing out America’s manufacturing base” is a discussion not had often enough. https://t.co/5hhk8RHWZl - **LambrosPetrou** (2025-04-13): https://x.com/LambrosPetrou/status/1911434459758448891 - @jachands @elithrar @threepointone I mentioned it in Step 3: https://t.co/TjoElcA69s I do think we need to move some of the patterns inside the main DO docs though, indeed. - **invisal89** (2025-04-13): https://x.com/invisal89/status/1911244334323896493 - I noticed a lot of love for diving into sqlite's internal from my last post. so I'm leveling up my game. Coming soon! #sqlite https://t.co/Ue7ifGxRt7 - **kimmaicutler** (2025-04-13): https://x.com/kimmaicutler/status/1911207762924843051 - Just when we were starting to make progress, we put a person in charge of California’s Senate housing committee, @aishabbwahab, who is poised to Everything Bagel every housing bill into total ineffectiveness. https://t.co/mnuv2MhosC - **kevinkern** (2025-04-12): https://x.com/kevinkern/status/1911195494493171774 - Cloudflare is all over my timeline, and ngl, they have a lot to offer. They’ve seriously improved their DX. I've just tried their remote MCP server example, which includes auth and a one-command deploy. Works. https://t.co/7XGzntQ8VX - **franpromano** (2025-04-12): https://x.com/franpromano/status/1911194586363339230 - @dok2001 @Cloudflare - improve the dashboard, the ux is terrible - add swr cache control header support - **theo** (2025-04-12): https://x.com/theo/status/1911164122286399559 - Over the last few months, I’ve noticed a huge vibe shift. Almost all of the best devs I know are suddenly embracing AI dev tools https://t.co/hOZmnm5M8Z - **dok2001** (2025-04-12): https://x.com/dok2001/status/1911148647510589677 - @kentcdodds @threepointone 🍿 - **PA80339646** (2025-04-12): https://x.com/PA80339646/status/1911140644673691929 - @dok2001 @Cloudflare node:fs support at durable objects level for when a DO is not enough but a container is too much. - **PA80339646** (2025-04-12): https://x.com/PA80339646/status/1911139118626619576 - @dok2001 @Cloudflare Full-stack(DOs, workflows, etc) developer infra in more locations. E.g., Brazil. Try measuring the latency of from end-user to DO off Brazil. You often end-up in Ashburn, VA, not even Miami. - **LN_onthewall** (2025-04-12): https://x.com/LN_onthewall/status/1911135069265068342 - @dok2001 @Cloudflare An MCP server for AI app builders to easily deploy all of cloudlfares offerings? - **dok2001** (2025-04-12): https://x.com/dok2001/status/1911134621237989681 - @CherryJimbo @Cloudflare @celso has been begging me to greenlight - **CherryJimbo** (2025-04-12): https://x.com/CherryJimbo/status/1911134104386490492 - @dok2001 @Cloudflare Solid email services would be awesome to see from Cloudflare. Worrying about delivery rates, spam, marketing vs transactional, different providers, pricing, etc. should be a thing of the past, like y'all made all the hurdles of SSL. - **dok2001** (2025-04-12): https://x.com/dok2001/status/1911132786162880993 - What should we build (or improve) next @Cloudflare? - **dok2001** (2025-04-12): https://x.com/dok2001/status/1911132219197845774 - Doing things that don’t scale can make a big difference. - **CloudflareDev** (2025-04-12): https://x.com/CloudflareDev/status/1911115347568304187 - You should be following @rohinlohe if you want to keep up with observability x Workers 👇 https://t.co/Fqlru2xAEe - **ankrgyl** (2025-04-12): https://x.com/ankrgyl/status/1911113217147044114 - Agent architectures are getting simpler - **threepointone** (2025-04-12): https://x.com/threepointone/status/1911104977017151996 - @kentcdodds hell yeah let’s do it - **kentcdodds** (2025-04-12): https://x.com/kentcdodds/status/1911099579686379591 - @threepointone would you like to do a live stream with me next week where you answer all my questions about https://t.co/3b7tKXBoNb, durable objects, and MCPs? - **threepointone** (2025-04-12): https://x.com/threepointone/status/1911070235261534367 - https://t.co/VzPFZA9MND https://t.co/xVNoLmkPFv - **zeeg** (2025-04-12): https://x.com/zeeg/status/1911056715942117440 - Whipped up a quick attempt at compatibility between vitest-evals and autoevals. If anyones using Braintrust (or autoevals) today, would love your feedback! - **thdxr** (2025-04-12): https://x.com/thdxr/status/1910952181588435312 - i am gonna keep talking about this because it keeps blowing me away - the more mature your codebase the more exponentially good vibe coding gets everyone stuck focusing on 0 -> 1 but being able to one shot new pages with perfect ui is outrageous - **pbteja1998** (2025-04-12): https://x.com/pbteja1998/status/1910944692939731320 - Working on launching the new version of https://t.co/NrSEJwMpCz It's almost ready for launch... People will soon be able to very easily create and share MCP servers with others. https://t.co/VPMIoS3FJk - **dexhorthy** (2025-04-11): https://x.com/dexhorthy/status/1910841628794773836 - "dex there are people out there building registries with thousands of MCP tools, why focus on just slack/email/chat integrations" 1) because serving an agent over slack is 10-100x more complex than just sending a slack message. but also 2) this is the whole game! https://t.co/fBRFczOdxk - **pitdesi** (2025-04-11): https://x.com/pitdesi/status/1910840885782130940 - @typesfast Btw any hoteliers reading this should check out https://t.co/fM2sbPEhbM which offers a beautiful modern booking experience, increases revenue and decreases fraud while taking payment online - **tnm** (2025-04-11): https://x.com/tnm/status/1910786626844057881 - MCP hackathon tomorrow in SF! We're sold-out, but have a waitlist: https://t.co/mAuYSJJYHp Will be highlighting the winners after! - **cramforce** (2025-04-11): https://x.com/cramforce/status/1910688106233421842 - Feature flags are about increasing velocity in 2 primary ways: - avoiding the need for long lived branches. If the code is ready, ship it and move on - allowing you to ship ASAP and quickly revert if there is a problem You can use FFs for experimentation but that's a choice https://t.co/4WNGYRvEqh - **GithubProjects** (2025-04-11): https://x.com/GithubProjects/status/1910533317172015227 - Turn any GitHub repo into a remote MCP endpoint - no config, no friction. https://t.co/miCTKxV4ds - **a85** (2025-04-10): https://x.com/a85/status/1910480803013795856 - Model Context Protocol (MCP) has quickly emerged as the preferred way for agents to connect to the outside world. Excited to announce that MCP support is coming to Postman! What does this mean: 1. Work with MCP requests natively in Postman. 2. Support for SSE and stdio transports 3. Seamless, optimized UI to reduce friction in working with tools. Reply to this post for early access! - **zebassembly** (2025-04-10): https://x.com/zebassembly/status/1910425006284419253 - We launched this yesterday and literally the day after I used it to root cause an issue. Building Cloudflare on Cloudflare to debug the issues we have while building Cloudflare on Cloudflare. https://t.co/QUL1XE1bJ7 - **boristane** (2025-04-10): https://x.com/boristane/status/1910423644225884255 - if you look at @CloudflareDev developer week, and you ask yourself "how do they do it?" the answer is simple: get smart folks and let them cook https://t.co/IUn1E930nk - **ID_AA_Carmack** (2025-04-10): https://x.com/ID_AA_Carmack/status/1910351545658466794 - I have never seen it expressed exactly like that, but I wholeheartedly endorse it: Feedback beats planning. My plea at Meta was “No grand plans, follow the gradient of user value”. https://t.co/LVrVLWBpe9 - **carlsverre** (2025-04-08): https://x.com/carlsverre/status/1909727930839974037 - Thanks for highlighting Graft @simonw! If you don't already follow Simon I highly recommend you fix that. They work on open source tools like Datasette for data journalism. https://t.co/m1LDTaMOot - **simonw** (2025-04-08): https://x.com/simonw/status/1909683896637370387 - Graft is a really interesting new open source replication system designed for read-write multi-leader replication, with a SQLite extension as the first demonstration of what it can do https://t.co/t4XZfaveQM - **backpinelabs** (2025-03-29): https://x.com/backpinelabs/status/1906013201692815845 - Roll your own @excalidraw backend using @Cloudflare Durable Object's websocket API - https://t.co/3jTLlAkwil - **pbteja1998** (2025-03-15): https://x.com/pbteja1998/status/1900869997523767540 - Wanted to learn more about MCPs this weekend. So I built this mini-tool called https://t.co/j8L28yGez9 Here is how it works: 1️⃣ Tell the AI what you want. 2️⃣ It creates an MCP server with various tools. 3️⃣ Talk with AI to add/remove/modify these tools. 4️⃣ Click on Deploy and get your own MCP Server URL. You can directly use in url in Cursor / Claude Desktop or anywhere you want. Completely powered by @CloudflareDev Durable Objects + Workers for Platforms - **Padday** (2025-02-25): https://x.com/Padday/status/1894424569542271240 - Slack doesn't matter. Docs don't matter. Sheets don't matter. Roadmaps don't matter. Strategies don't matter. Shipping code to real customers and watching that code getting used is all that matters. That needs to be the obsession. Everything else is a means to an end, and without running code is ultimately valueless. - **anthonyVslater** (2024-12-29): https://x.com/anthonyVslater/status/1873235520740049294 - Microphones caught Draymond Green lighting into Buddy Hield tonight (“Wake the f*** up or go sit the f*** down”). He went into detail about that moment and the general leadership this team needs. “Mics catch everything today. I don’t care. I’ll say it right into the mic.” https://t.co/phIqnncLfo - **burcs** (2024-12-19): https://x.com/burcs/status/1869810528786501801 - You might be thinking to yourself, I don't really want to write code on my phone, yeah me neither... Coming Soon™ https://t.co/BRiGn53lz9 - **kdy1dev** (2024-12-08): https://x.com/kdy1dev/status/1865719533920755829 - Is there a library for node.js that profiles total CPU usage of a function call? I want it to track the total execution time excluding the time spent waiting for other code or being blocked by system calls - **Conor_Harris_** (2024-11-05): https://x.com/Conor_Harris_/status/1853918909956780474 - This is a big bang-for-your-buck exercises that addresses the needs of both sides in lateral pelvic tilt & uneven hips! https://t.co/trBODFv1Ll - **pzakin** (2024-10-04): https://x.com/pzakin/status/1842255551117332514 - //open source saas and speculative thoughts on post-scarcity software Assume the cost of building software goes to zero. One view is this is a death knell for SaaS. An alternative view: if software becomes extremely cheap to make, companies stop selling software and start selling agents. They use or build open source platforms and monetize purely with an agent. E.g. next-gen zendesk is just open source zendesk that sells a competitive agent product. Competition at the agent layer is complicated. OpenAI will compete here. Can wrappers truly outcompete an OpenAI digital worker post-AGI? I'm not sure... But my sense is that as cost of building business software goes down and appetite for extensibility increases, open source SaaS is structured in a way that anticipates long-term competitive dynamics. If you're building open source business systems, I'd love to chat... - **threepointone** (2024-09-27): https://x.com/threepointone/status/1839588924446159278 - One more thing... - **arjunkhemani** (2024-09-22): https://x.com/arjunkhemani/status/1837704603334226189 - .@naval: If I look back on my life, almost everything great that I managed to pull off—great by my own definition, not by the world’s definition—has come from following my own natural intellectual obsessions. So I think if you can get obsessed over something, and if you can dive into it, just let yourself go and learn everything about it with no motivation other than just wanting to know the answer, that becomes the basis for all of the so-called “self-improvement” out there. You have @bryan_johnson here somewhere. He’s obsessed with not dying and aging. Well, he’s obsessed, and that’s great. He’s following his intellectual obsession, and we all get to learn from that. You’re following your obsession with the network state, and there are people out there following their obsession on AI, on crypto, and on history—Roman history, whatever it is. But if you get obsessed with something, you can figure it out to a detail that other people don’t. You can satisfy yourself. If you go deep enough into anything, you find the same commonalities, you find the same philosophies. As a weird aside, I’ve gotten into photography recently. Don’t ask me why, but it’s just a way of combining art and science, being social and antisocial, doing something utilitarian. But sure enough, I’m obsessed with photography. I’m reading all the philosophical photography blogs where the authors talk about the meaning of life, art, science, and so on. If you go deep into anything, you’ll find the same common threads. And so, I think self-improvement really just comes from letting yourself be who you are—following the things you truly want to follow, figuring out what you want to figure out—not worrying about what others want or think. And then, you kind of find yourself in the same place at the end, no matter which route you take. - **LambrosPetrou** (2024-09-02): https://x.com/LambrosPetrou/status/1830595556839707107 - Datadog Continuous Profiler is just amazing 👌🚀 If you are using DDOG, but not profiling, just go enable it, now! https://t.co/WxbVZiE1l9 - **cejapower** (2024-04-24): https://x.com/cejapower/status/1783243484863848736 - CA is decades behind other states in terms of constitutional environmental rights. #ACA16 would enshrine everyone's right to clean air, water, and a healthy environment into our constitution - we hope to see it on the ballot in Nov. 🙏🏽 @ib2_real https://t.co/RqS1hJ7nnN - **schickling** (2024-04-21): https://x.com/schickling/status/1782072068454150564 - What are currently the biggest limitations when using Cloudflare Workers for server use cases e.g. compared to a Node.js server? - **KentonVarda** (2024-04-05): https://x.com/KentonVarda/status/1776236078749335654 - For more details and a bunch of examples, check out the blog! https://t.co/qHsBwlBlRO - **martin_casado** (2024-03-01): https://x.com/martin_casado/status/1763654159516254517 - Come for the model. Stay for the workflow. - **mipsytipsy** (2024-02-12): https://x.com/mipsytipsy/status/1757166124339437714 - "Every infrastructure decision I endorse or regret after 4 years at a startup" -- https://t.co/4DMDKz2vtT Best infrastructure post I've read in YEARS; would love to see this become a hot series. Write one about your choices, please!!! - **rauchg** (2024-02-11): https://x.com/rauchg/status/1756721008529260815 - https://t.co/HOvrPGxznA - **anthonyVslater** (2024-02-11): https://x.com/anthonyVslater/status/1756546095612891173 - Draymond Green on the ‘too small’ back-and-forth taunts with Jusuf Nurkic: “You can’t be a nothing defender if you gonna do that.” https://t.co/va7P9w9MpY - **kaslinfields** (2022-05-18): https://x.com/kaslinfields/status/1526910075381612545 - When should you use MIGs vs timesharing to optimize your GPU utilization? Timesharing is a logical sharing of the GPU, whereas MIG is a physical partitioning of the GPU- so there are higher hardware considerations. MIG is more limited in the configurations it supports. https://t.co/z4kn3Ojl3m - **justindross** (2021-02-03): https://x.com/justindross/status/1356797946805833731 - Jeff Bezos’s letters are better than any MBA He’ll teach everything you need to know about developing high standards, and it doesn’t even take him two pages https://t.co/UDVs3W1GhT - **zachtratar** (2021-01-23): https://x.com/zachtratar/status/1353064295785877507 - Love how these kitchens mix a futuristic feel w/ raw materials. Goals. 🤩 https://t.co/J9HrzvyTAH - **alexandr_wang** (2020-11-26): https://x.com/alexandr_wang/status/1332059627878400000 - Breaking down my recent post: 1/ @scale_AI, I interviewed everyone we gave an offer to for a long time. I wrote a memo to the company about what I look for. I wanted to share it with the community because I don’t think people do this enough. Thread👇 https://t.co/Da5OFUATWT - **packyM** (2020-11-12): https://x.com/packyM/status/1327021201072459776 - Dear LazyTwitter, Someone made a thread on a survey they did about Slack vs. Microsoft Teams usage a month or so ago, I think showing that Slack was still most popular among startups & technical teams. Does anyone know what I'm talking about? $WORK Warmly, Packy - **karpathy** (2020-11-07): https://x.com/karpathy/status/1325154823856033793 - How to become expert at thing: 1 iteratively take on concrete projects and accomplish them depth wise, learning “on demand” (ie don’t learn bottom up breadth wise) 2 teach/summarize everything you learn in your own words 3 only compare yourself to younger you, never to others - **lessin** (2020-11-06): https://x.com/lessin/status/1324845846274347008 - Why you need to take risk (and leave your big-company job) if you can afford to… https://t.co/iiW1kIaQRl - **Negative_Tilt** (2020-08-19): https://x.com/Negative_Tilt/status/1296004308186480640 - #Timelapse: Sudden explosive development of the #CZUAugustLightningComplex #fire burning tonight in San Mateo & Santa Cruz Counties. @NWSBayArea @CALFIRECZU #CAwx https://t.co/pADZ3B4Trk - **energysmartwv** (2020-08-18): https://x.com/energysmartwv/status/1295840798844559361 - @Negawatt_Nate 4G Lapolla https://t.co/QdqAbm0rjn - **yinonw** (2020-08-17): https://x.com/yinonw/status/1295152606646394880 - So let's review: * No matter the intervention policy, hot spots decline when reaching ~15% antibody expression * Given a sufficiently large sample size, all hot spots stopped at 20-25% * Masks, businesses, schools, etc. all fail to explain it - **lawmaster** (2020-08-13): https://x.com/lawmaster/status/1293827154061533185 - Price of YAM went from nearly $15 to less than $0.7 in just a few minutes. Liquidity dried up really quickly - **lawmaster** (2020-08-13): https://x.com/lawmaster/status/1293823726962835456 - Ok now is time fellas https://t.co/6T3EXmT3bH https://t.co/MvgiYsRVhl - **NZClarke** (2020-08-08): https://x.com/NZClarke/status/1292200151734345728 - @pv_reynolds Paris car ownership has dropped to 15% so they are doing cool things with redundant underground car parks. I was taken on a tour, converted to hydroponics growing microherbs and fancy mushrooms for top end restaurants all delivered to order by bike https://t.co/8Tx2eIo79J - **GRoditiD** (2020-08-04): https://x.com/GRoditiD/status/1290469577621680128 - social platforms have not proven to be truly long-lasting. even all-mighty FB is on the decline if measured by the value of its userbase and the youths do not care for IG. maybe social platform user-bases should be valued like run-off assets that need to be constantly re-acquired https://t.co/auVFqIyvXM - **HuXijin_GT** (2020-07-29): https://x.com/HuXijin_GT/status/1288326960737738754 - Chinese netizens suggest the closed US Chengdu consulate be transformed for these purposes: hot-pot restaurant, Huawei flagship store, or exhibition of Chuanjianguo’s good deeds. Chuanjianguo is a nickname given to President Trump by Chinese netizens, meaning Trump builds China😀 https://t.co/e7Bikijwmq - **marketplunger1** (2020-07-22): https://x.com/marketplunger1/status/1285974128848232448 - Just finished reading Will O'Neil's book "How To Make Money in Stocks" Tobi's (@Greenbackd) podcast with @PaulAndreola is the next best thing to listen to after reading that book. Listen here: https://t.co/DEPLp1teBD Paul's thoughts on the CANSLIM method opened my eyes: https://t.co/RMGiWFReRw - **lessin** (2020-07-13): https://x.com/lessin/status/1282713197921828865 - Tesla and Amazon - crazily enough - are now equivalent 5 year bets... read in to that what you want. https://t.co/wJ752SKnrh - **sarahkliff** (2020-05-29): https://x.com/sarahkliff/status/1266408985437102085 - This chart is bonkers. This is health care spending — which has marched upwards for 50 years, even during recessions — falling off a cliff. https://t.co/sGJLnxs6vS - **tracyalloway** (2020-05-20): https://x.com/tracyalloway/status/1262901765415960576 - The composition of global fiscal stimulus in the corona crisis of 2020 vs. the financial crisis of 2008/2009, via UBS. Fiscal stimulus now amounts to 3.7% of world GDP: https://t.co/VzJQ8hD93K - **arrington** (2020-04-05): https://x.com/arrington/status/1246621315336822784 - This is not good. https://t.co/jnrPRR9qXc - **steobrien** (2019-10-23): https://x.com/steobrien/status/1186994151130308609 - Apple’s software development practice is stuck in the past. Great article from an 18-year veteran. https://t.co/vcg9Mv2t8k - **TheStalwart** (2018-04-13): https://x.com/TheStalwart/status/984798850228674560 - Finally! 12 years after I proposed it, the business model of startups shorting the stock of incumbent players (HT: @matt_levine and @pearkes) https://t.co/V5xUywzh7w https://t.co/ZUZ9EKnOIs - **sama** (2015-08-10): https://x.com/sama/status/630869228137127936 - The most impressive people I know spent their time with their head down getting shit done for a long, long time.