🤸🏽

@boristane @justinsunyt what have you been doing instead?

@justinsunyt came to the same conclusion and removed the kanban board I had in nominal.dev

kanban doesn’t make sense for coding agents we tried it 6 months ago. every task just ended up in the “needs review” column

@michellechen @ashishkmr472 is real mvp

trying this new cloudflare tool and i’m sold. 🦭🦭🦭 @dcarter_js cooked

build the tool, use the tool, and use the tool to build the tool. none of these are mutually exclusive

Give your agent the Stripe projects CLI and have it: - create a Cloudflare account - start a subscription - build and deploy apps every day getting closer to starting a new business being literally a one-shot prompt more soon! just the beginning…

want to get a deep dive on how we made artifacts, some novel use cases, and what we're cooking up for the future? check out this convo + overview w/ @craigsdennis and @elithrar cloudflare.tv/event/dg76WLWZ

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 blog.cloudflare.com/artifacts-git-…

You can now add voice to your agent using Agents SDK: blog.cloudflare.com/voice-agents/ Voice is just another input -- you can use the same WebSocket connection your Durable Object uses to transmit audio. So much fun working with @threepointone on this

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! blog.cloudflare.com/project-think/

Cloud 2.0. Built to scale from day 1. Deploy to region: earth.

Friday drop pre Agents Week — concurrent connection limits on Workers are now in practice a thing of the past Before: With 6 concurrent outbound connections from one request, new connections from your Worker were blocked until an earlier connection fully completed

really liking this project

@biwills @zerocipherz we’ve got some really exciting stuff here coming soon

100% I have this issue as well and seprate workers is the only solution I found. Breaking it up is the only way I've found to have different sets of secrets. Then I have ~600 lines of bash that break it into three groups: production, staging, and PR previews. This makes it easier to set things like the OAuth Proxy plugin for Better Auth, other dynamic vars, and also creating & seeding a test database from staging. But I agree with you @zerocipherz, it constantly feels like I'm either fighting the platform to do what I can do trivially in other PaaS and missing something obvious or doing something wrong. @dcarter_js is one D1 database / the same set of env vars shared across all environments what cloudflare workers is built for? I'm not sure I understand the usecase for that, but would be curious to hear what it is even if it's not my usecase 🙂

Why can't we deploy workers per branch and get a preview URL easily, similar to pages? I would love this feature. Right now i have to deploy separate workers to make this tenable so i can have proper environments. Am i doing something wrong? i could never get deploy previews working on 1 worker in the way I expect it to work. If we could connect the worker to multiple branches and get truly separate environments, that would increase DX drastically i think. cc: @dillon_mulroy @CloudflareDev

@dillon_mulroy @zerocipherz very interesting 👀

@zerocipherz cc: @dcarter_js

@KentonVarda drives me insane

How is Google so far behind on agentic AI? This is the Gemini sidebar *embedded inside gmail*.

@dillon_mulroy need

pi 🤝 raycast

🚀 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? 📷👀

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 ⤵️ blog.cloudflare.com/emdash-wordpre…

@balazsorban44 @vercel what a run - excited to see what’s next

🖤 After 4+ years, I've made the decision to leave @vercel. March will be my last month. It was a true character-shaping experience, that made me better understand priorities both in my career and life. I'm thankful for all my managers and everyone I've worked with!

One point of confusion I'm seeing: Lots of people assume you're supposed to run the AI agent harness itself inside a sandbox, and wonder how to do that with Dynamic Workers. The harness does *not* need to be in the sandbox. It can be a regular old Worker (probably a Durable Object, probably using Cloudflare Agents SDK[0]). Remember that an LLM can't manipulate the world at all unless you give it tools. So if you are careful about what tools you give it, it is inherently sandboxed. What we are proposing here is that you basically give the agent two tools: One to explore the APIs it has available to it (returning TypeScript type definitions), and one to execute code against those APIs (input is JavaScript). The latter tool runs the code in a Dynamic Isolate. You can't quite just run Claude Code or Codex in there, since they aren't designed to run in Workers, at least today. But @southpolesteve managed to get OpenCode (a similar harness) running 100% in Workers[1]. It's also not as hard as you might think to write your own harness -- I've done it. That said, if your agent's task specifically involves using a bunch of Linux programs, it may still be best to use containers for that. For instance, if your agent is doing software engineering -- checking out a git repo, running a compiler, etc. -- it may be impractical to make that all work in Workers. Try our Sandbox SDK[2] for that. But if your agent is planning a family vacation, it probably doesn't need to run rustc. It probably just needs to interact with a bunch of APIs. And that's where Dynamic Workers shine. blog.cloudflare.com/dynamic-worker… [0] developers.cloudflare.com/agents/ [1] x.com/southpolesteve… [2] developers.cloudflare.com/sandbox/

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: developers.cloudflare.com/dynamic-worker…