The team is plain text: job descriptions, a process, automatic checks and one shared record. That's why it works with whichever AI tool you already use, and why you can read every rule yourself.
One job description per agent: what it does, what it never does, who checks it.
The process: engagement, research to spec, foundation, build, production, improve.
How every handoff and verdict is written on GitHub, so any agent can pick up the work.
How the team plugs into each AI tool, and how each role's tier maps to a model.
The method, and the research behind each choice.