Mention @aizu on any issue or pull request and it spins up an isolated coding agent to handle the request — commits, pushes, and creates PRs autonomously.
- Run any model, own your infrastructure. Supports pi.dev and any local model server (e.g. llama.cpp, Ollama) for zero recurring cost.
- Self-hosted, no telemetry, no accounts. Own your data and run anywhere Docker runs.
- GitHub-native and fully autonomous. Comment on issues and PRs like you would with a teammate. Agents run headless — no human-in-the-loop until you review the PR.
@aizu fix this
@aizu add error handling for the empty list case
See the Quickstart guide for the full setup.
Aizu is built for agentic engineering — the practice of experienced engineers using autonomous agents to amplify their expertise. Agents that can generate, execute, and iterate on code independently between human checkpoints aren't replacing developers; they're making them faster. The key is keeping the human in the loop at the right level: reviewing PRs, not babysitting prompts.
The project is intentionally simple. The entire service is a webhook handler, a Redis queue, and a container orchestrator. The actual intelligence comes from the agent harness you use — pi.dev by default, or any other tool that runs in a container. Aizu just connects it to GitHub.
You own the whole stack. There's no hosted service, no telemetry, no account to create. Fork it, modify it, run it on a Raspberry Pi or a GPU cluster. Point workers at a cloud Redis and scale horizontally, or run everything on a single machine with docker compose up. See the full documentation for technical details.
MIT