The right model for each role.
Every agent picks its own CLI and model. Plan and review with your strongest model, and code with a faster one, all on one board.
Your PM agent drafts the plan, and you approve it. Specialist agents build in parallel. Nothing is done until you accept the diff.

Open source (Apache 2.0). Runs locally, no Coppice account needed.
Hand off the work. Write a ticket, or approve one an agent suggests. Assign it to your PM agent, which drafts a plan you approve before anyone writes code.
Let agents build. Each ticket gets its own git worktree, so parallel agents never touch your main checkout. Follow any run live.
Human Review. Read the diff in the app, comment on any line, and nothing counts as done until you accept it.
Download, open, work. Nothing to set up: no Docker, no Coppice account.
Every agent picks its own CLI and model. Plan and review with your strongest model, and code with a faster one, all on one board.
Every agent's work is a card on one board. Blocked tickets and work waiting for your review stand out, so you know where you're needed.
Every change waits for you as a diff in the app. Nothing counts as done until you accept it, and what you accept is exactly what merges.
Coppice drives the CLIs you already pay for: Claude Code, Codex, Cursor, OpenCode and Kilo Code.