$200 FREE CREDIT! We just launched our inference platform for beta testing, and we're giving it to the community first.
⭐ Star SGLang on GitHub (https://t.co/uEeiF4ANRf) + repost this to claim your credits.
→ Limited spots, first come first serve
→ Deadline: May 13, 2025 (AoE)
Every star, every issue filed, every PR reviewed, every question answered in Slack — You built this with us. Thank you for believing in open-source AI infrastructure, in our mission, and in us.
Claim your credits: https://t.co/MVDvcvkFGX
@thdxr@UpToNathan Nice. And 1 more benefit is that maybe it can run on Android termux. Currently bun cannot run on termux so any Bun API based thing cannot run.
I built Chrome-ACP, a Chrome extension that lets AI agents control your browser.
Works with Claude Code and any ACP-compatible agent.
Also, like #OpenClaw , it can be deployed to anywhere and just use phone to control your agent.
https://t.co/9mztjXznIy
@devongovett It is more accurate now. Most of the code works, and human is still needed to review. All your reasons can be avoided if the method is right. I no longer write code but keep reviewing code generated by AI, and things work pretty good.
@shadcn@bunjavascript Super convenient!!!
Bun v1.3 introduces `build --compile` to compile a full stack app into a single executable. But currently it does not work well with the shadcn template.
I fixed that with a PR. @shadcn@bunjavascript
https://t.co/eiKANIGnxO
Bringing A2A and Human-In-The-Loop to live.
I just built a web UI of A2A with Human-In-The-Loop feature.
Everyone can try it on:
https://t.co/OvpOh0IVGR
Here is the repo:
https://t.co/uuXws9szCl
Built with A2A, ai sdk, ai elements, next.js.
#A2A#HITL
#A2A is good, but has anyone adopted it yet?
There are a few A2A examples in the community, but I haven't found one with #HITL (Human-In-The-Loop) feature.
So I just built one, posting it here for everyone to correct me if I'm wrong.
https://t.co/uuXws9szCl
Here’s the app idea. But first the problem:
- Our conversations are scattered across apps
- Answers are wall of text. You only care about 1-2 lines
- Copy-pasting useful parts is tedious
- And good luck finding where you ask what
Coppermind is a small...(continues in thread).
Announcing build-my-own, an MCP Server that helps to set up suitable folders and AI rules to turn Agents like Cursor @cursor_ai and claude code into the best tutors who instruct you to replicate any github project from 0 to 1.
Any feedback is welcome!
https://t.co/goqWV9HMdh
Started to use Zed @zeddotdev today. Noticed it years ago, and today I found it smooth and acceptable for me.
But I miss the HTML snippet from Vscode, and had to configure it myself in Zed. At last everything works fine!
Post a gist that helped today.
https://t.co/e0NZWLsOC5
My PR to MCP TypeScript SDK got merged. But used company git account for commits, so I guess my personal GitHub still shows zero commit contributions. Kinda bittersweet.😅
avoiding dystopia? imho open-source, widely distributed AI is the only way
Supergateway v3.2 for MCP is live. 1.6k GitHub stars. 130k installs
contribute, star us, help open-source grow
thanks to @ollogllg for this release!
run Streamable HTTP in any MCP client
Supergateway v3 release just went out. whats in it:
- Streamable HTTP to stdio support (community heroes did it)
- updated MCP sdk
- more reliability due to proper tests