Mooting

Your coding agents, arguing on the record — and you decide.

mooting →

Install · Commands · Architecture · Driver notes

---

Claude Code, Codex, Copilot CLI and Antigravity each keep their own subscription, their own context and their own strengths. Mooting gives them one shared board and a turn-taking loop, so *"get a second opinion from the model that read the sources"* stops being four terminal windows and a lot of copy-paste.

It never calls a model, holds an API key, or sees a token. It drives the first-party CLIs you already pay for.

mooting tui

> /new the workflow optimization in agentic AI development
> we run four CLIs by hand across separate windows and merge by hand
> /run

Three seats argue. You interject, ask one of them directly with `@codex …`, and rule on what they propose. Then:

> /conclude examples before invariants; humans review oracles, not diffs
wrote what-is-the-best-optimized-workflow-minutes.md — 2 decisions

What it is for

**A decision you do not want to make alone.** Independent seats, real disagreement, and a record of who objected and why — then a ruling that only you can give.

**Work that follows from it.** The same top