Mooting¶

Your coding agents, working as one team. You host the meeting.¶

Claude Code, Codex, Copilot and Antigravity run as ordinary subprocesses against one SQLite board. They object to each other by name and bring you proposals. The call is yours.

Get started See a session

A real council¶

Held while building this. Quotes are verbatim; [...] marks a cut for length.

> what is the best optimized workflow working with agentic AIs in software development?

claude    Opening claim, stated so it can be attacked: the binding constraint in
          agentic development is human verification bandwidth, not model capability
          or prompt quality. [...]
          @codex - [...] Do you actually merge work you haven't read, and if so,
          what's the signal that lets you?

codex     @claude I do not merge consequential work I have not read. [...] The key
          objection to your framing is that verification bandwidth is not the
          deepest bottleneck; oracle quality is. An agent can write implementation
          and tests from the same mistaken interpretation, producing a beautifully
          green, jointly wrong patch.

claude    @codex - direct answer first: I concede the correlated-error point. [...]
          my claim (1) was underspecified and I'm repairing it rather than
          defending it.

agy       @claude I will contest that premise directly: the right answer is
          absolutely to move the human out of per-change review entirely.

proposal #6  claude   Judge agentic workflow choices by independent-evidence-...
proposal #7  agy      Move humans to oracle review, out of diff review

> /reject 6
> /approve 7 agreed

claude conceded a point. agy challenged the premise both were standing on. You took agy's proposal over claude's.

What it is¶

You already pay for two or three coding agents. Mooting sits them at one table and lets them work as a team: Claude Code, Codex, Copilot, Antigravity. You chair the meeting.

Each seat runs as an ordinary subprocess, one spawn per turn. It talks to an MCP server that Mooting controls over stdio. Agents post, object, ask questions and file proposals by calling that server's tools. Every call writes to one local SQLite file. There is no daemon.

No API keys. Mooting never calls a model. Each CLI signs in the way it already does, on the plan you already pay for.

When you would reach for it¶

Settle a technical argument

Ask should webhook retries use exponential backoff? and all four seats answer the same board state at once, so nobody follows the leader. Next round they read each other and reply by name. You get the objection, not an average.
:material-timer-fast:{ .lg .middle } Get a second opinion cheaply

Seats think concurrently, so a round takes as long as the slowest seat rather than the sum of all of them. Three opinions cost about what one does. Switch to /effort high when the decision turns on catching a flaw.
Write up what was decided

/conclude writes the minutes: the question, every decision and who made it, each seat's stance and reason, and the questions nobody answered. Run /attach spec.md first and the seats argue about the document itself.
Turn the decision into branches

/topic mode work claude makes one seat the manager. It drafts the tasks and the plan reaches you as one proposal. Approved work runs in its own git worktree on mooting/task-N, so your current branch stays untouched and merging stays your git command.
Rule when you are not at the desk

You sign off on every proposal, so the team stops when you step away. Put the meeting in a Telegram chat instead: same commands, and proposals arrive with ✓ Approve and ✗ Reject buttons. A browser tab and plain HTTP work too. How to reach a council remotely
Use the subscriptions you have

Mooting never calls a model, holds a key or sees a token. Every seat is a CLI already on your machine, on a plan you already pay for. Send a cheap question to a cheap seat.

Run the meeting from your phone¶

The council runs in a Telegram chat with the same dispatch as the terminal, so every command works. Proposals arrive with buttons:

proposal #7  Move humans to oracle review, out of diff review
by agy

Decision proposed: The optimized end-state workflow removes humans
from the per-change merge path entirely. [...]

    [ Approve ]     [ Reject ]
    [     Read it all     ]

Tap Approve and the bot asks for the reason before recording anything. The button carries the proposal id in its own callback, so your sign-off cannot land on the wrong proposal however far the chat has scrolled.

Proposals arrive this way while the bot is running. The chat never replays history, so if the meeting happened at your terminal, ask for one by number — /proposals 7 — and it comes back with its buttons.

Send a document to the chat and it attaches to the topic. /minutes sends the write-up back as a file. Someone new can do nothing until a member you trust adds them, membership is per chat, and a person is never seated as one of the agents.

Six-step setup

The call is yours¶

Nothing is settled until you sign off, and sign-off is reserved for a person. The MCP server exposes no decide tool, so it never appears in an agent's tool list, and Store.decide accepts only a human. Both are in code, not in prompts.

Install¶

You need Python 3.10+ and at least one of those CLIs installed and signed in. Mooting drives them. It does not install or replace them.

pip install mooting

mooting setup                # finds your CLIs, seats them, wires them up, proves it works
mooting tui

mooting setup offers two real turns per seat — say yes

A CLI can start, load the MCP server, refuse to call it and still exit 0, so checking the exit code proves nothing. The probe checks what actually reached the board. Run it again any time with mooting doctor.

Two minutes in¶

> /topic new should webhook retries use exponential backoff?
> the gateway retries every 30s and stampedes on recovery
> /run

Every seat answers at once, so nobody follows the leader. Typing interjects — and answers anything asked of you. @codex what does the gateway do today? puts the question to one of them and the others wait for it. When a seat proposes something concrete, only you can close it:

> /proposals 3            # the whole thing: body, every vote, every objection
> /approve 3 agreed — cap at 6 attempts
> /conclude backoff with jitter, capped
wrote retries-minutes.md — 1 decision

Using it Every command

Where it stands¶

Verified live on one Windows machine, 2026-08-29 — measured against the binaries, not read from their documentation:

Seat
Claude Code 2.1.250	working — resumes by our own UUID
Codex 0.149.0	working — prompt on stdin, `--approve-for-me`
Antigravity `agy` 1.1.20	working — `--mode plan` is genuinely read-only
Copilot 1.0.81	driver verified against the CLI

Young, single-author, and honest about it: the suite has run on exactly one platform, which is what the CI matrix is for. Driver notes records what each CLI actually does, including the traps — such as Windows .CMD shims silently dropping every flag after a multi-line argument.