Brain is an agent runtime that allows you to build AI native system that runs tools on anywhere from browser to sandbox; fully customizable so you can choose to run on pi/codex backed agent-loop (securely in wasm-based isolation). Its intentionally minimal and allow you to build agent beyond coding, it comes with real-time events so you could tell instantly what's going on.
I built it because I realise how complicated and tedious agent platform it had become. I want to create a minimal platform where people can write extensions to customise every aspects of the runtime.
This project is heavily inspired by philosophy of Pi coding agent (https://pi.dev/) and Anthropic’s engineering blog on decoupling the brain from the hands (https://www.anthropic.com/engine...)
I want to build something people want! Let me know how bad/good it is and why you would/would not use it! Cut the noise and reach me directly at luoweilue@gmail.com
Hello Hackers!
I’m Weilue Luo, the creator of Brain.
Brain is an agent runtime that allows you to build AI native system that runs tools on anywhere from browser to sandbox; fully customizable so you can choose to run on pi/codex backed agent-loop (securely in wasm-based isolation). Its intentionally minimal and allow you to build agent beyond coding, it comes with real-time events so you could tell instantly what's going on.
I built it because I realise how complicated and tedious agent platform it had become. I want to create a minimal platform where people can write extensions to customise every aspects of the runtime.
This project is heavily inspired by philosophy of Pi coding agent (https://pi.dev/) and Anthropic’s engineering blog on decoupling the brain from the hands (https://www.anthropic.com/engine...)
I want to build something people want! Let me know how bad/good it is and why you would/would not use it! Cut the noise and reach me directly at luoweilue@gmail.com
This is cool. I would like to see it compared to Pi, OpenCode, DSH in your benchmarks.