Fully autonomous AI Agent that can perform complicated tasks and projects using terminal, browser, and editor.
Discord: https://discord.gg/uMaGSHNjzc
- 🔓 Secure. Everything is running in a sandboxed Docker environment.
- 🤖 Autonomous. Automatically detects the next step and performs it.
- 🔍 Built-in browser. Fetches latest information from the web (tutorials, docs, etc.) if needed.
- 📙 Built-in text editor. View all the modified files right in your browser.
- 🧠 All the history commands and outputs are saved in the PostgreSQL database.
- 📦 Automatic Docker-image picker based on the user task.
- 🤳 Self-hosted
- 💅 Modern UI
- golang
- nodejs
- docker
Add to .env
file in the backend
folder.
OPEN_AI_KEY
- OpenAI API keyDATABASE_URL
- PostgreSQL database URL (eg.postgres://user:password@localhost:5432/database
)DOCKER_HOST
- Docker SDK API (eg.DOCKER_HOST=unix:///Users/<my-user>/Library/Containers/com.docker.docker/Data/docker.raw.sock
) more info
Optional:
OPEN_AI_MODEL
- OpenAI model (default:gpt-4-0125-preview
). The list of supported OpenAI models can be found here.
Frontend environment variables can be set by creating a .env.local
file in the frontend
folder.
VITE_API_URL
- Backend API URL. Omit the URL scheme (e.g.,localhost:8080
NOThttp://localhost:8080
).
- Run
go run .
inbackend
folder - Run
yarn dev
infrontend
folder - Open your browser and enjoy!
- Agent API
- Frontend
- Backend API + PostgreSQL integration
- Docker runner
- Terminal output streaming
- Browser output streaming (in progress)
- Editor output
- SWE-bench
- Better way to run it (eg a single docker command)
This project wouldn't be possible without: