How it works

my-ai-team turns a backlog of GitHub issues into reviewed and merged pull requests while you set the direction.

The delivery loop

You make one handoff; the steps after it run without you.

1. You mark an issue ready. That label is the whole handoff. Work passes through GitHub issues and labels: claiming a ticket swaps ready for a lock label naming the mode.
2. An agent claims the ticket. A launch selects the next ready issue and takes it, unless you name an issue directly on the turn.
3. The agent works in its own worktree. Every worktree-creating mode runs in a dedicated worktree or a read-only snapshot, isolated from your main checkout, so unrelated or uncommitted work stays out of the change.
4. The agent implements the change and opens a pull request. You see the pull request like any other.
5. A separate reviewer checks the work, and the result is merged. In duo and team, an independent reviewer gates the plan and the implementation before the merge.
6. You are asked only at decisions. Notifications and questions reach you, and you can answer them from your phone over the optional Telegram relay: every agent message arrives in Telegram, and your reply routes back to the exact agent that spoke.

What you need

• A GitHub repository.
• One supported backend CLI: claude, codex, copilot, pi, opencode, freebuff, or cmd (Command Code). You bring your own account and token or login; my-ai-team runs on top of that CLI and does not replace it.
• Linux, macOS, WSL2, or Windows Git Bash.

Ready to run it? Follow the Linux quickstart or any of the other quickstarts under the product overview.