Agent2077
A self-hosted AI agent workspace with a browser UI that connects to local or OpenAI-compatible models and can edit code, run shell commands, deploy Docker-backed apps, search the web, and maintain memory across sessions on your own Linux machine.
Agent2077 runs on a Linux machine and serves a full browser UI reachable from any device on the network, connecting to any OpenAI-compatible model endpoint such as LM Studio, OpenRouter or a remote vLLM server. The agent can chat with web search, open a project and read, write and edit code across its file tree, execute shell commands and run tests, and build and containerize web apps or games with Docker to serve on the local network.
It stores facts and reusable skill instructions and recalls context across sessions, supports multi-model orchestration, and includes a Self-Dev mode in which the agent can improve and modify its own codebase. Everything runs on the user's own hardware, with no cloud account required beyond whatever model endpoint is chosen.
Built with TypeScript; an install script sets up Docker, Node.js 22 through nvm, nginx, Avahi mDNS, a local SearXNG search instance and a systemd service so it starts on boot. AGPL-3.0 licensed.