my-ai-team documentation

Turn a backlog of GitHub issues into reviewed and merged pull requests — you set the direction and take the decisions.

This site reflects the latest release (v4.72.0) only. Historical release changes are documented on the Changelog page.

How it works

1. Label an issue ready

Tag a GitHub issue as ready in your backlog. That is the whole handoff.

2. An agent claims it and opens a pull request

An agent picks the issue up, plans the change, implements it, and reviews its own work before opening a pull request for a separate reviewer.

3. Review and merge happen without you

A separate reviewer checks the work and merges it. Decisions that need you reach you, optionally over Telegram.

Solo maintainers with a long backlog

Agents claim your ready issues, implement them, and merge the results while you set the direction.

Small teams that want async delivery

Drive and intervene remotely over Telegram instead of watching an interactive terminal.

Anyone already paying for a backend CLI

my-ai-team runs on top of the CLI you already pay for — it does not replace it, and you keep using your own account.

Understand the modes

Seven modes cover the whole loop, from one agent working alone to a planner, developer, and reviewer relay. Read the modes.

Compare with hand-run agents

See how a managed backlog loop differs from running a coding CLI or an IDE agent yourself. Compare.