MootDeck
Get notified

Docs

MootDeck, without the fog.

A local desktop canvas for real CLI coding agents needs more than a landing page. These notes explain what runs locally, what the ropes gate, what can leave the machine, and what is still rough.

Blueprint-style diagram of terminal nodes connected by rope paths on a MootDeck canvas.
Terminal nodes, ropes and notes — the moving parts these docs walk through.

01What this is

MootDeck needs explanation, not fog.

MootDeck is young, closed source for now, and built around a permission model people will question. That is fine. Serious developer tools do not win trust by sounding finished. They win it by showing the edges.

These pages are written for the person who likes the idea, then immediately asks the uncomfortable questions: what runs locally, what reaches the network, what does a rope actually gate, and where does the beta still bite?

02Shape

What MootDeck is, and what it is not.

It is

  • A local desktop app.
  • A canvas for real terminal sessions.
  • A small CLI for roped agents: list, check, ask, recruit, note, and connect.
  • A way to make multi-agent coordination visible.

It is not

  • Not an AI model.
  • Not a token proxy.
  • Not a cloud agent platform.
  • Not a replacement for Claude Code, Codex, aider, or your shell.
  • Not an autonomous software factory. That sentence can die in a ditch.

03Reading path

Start where your doubt is.

02

How it works

The concrete loop: start terminals, draw ropes, recruit agents, pass questions, write notes.

 03

Security model

One-hop, non-transitive command visibility, IPC auth, local storage, and the egress matrix.

 04

Known limitations

Closed source, beta UX, Windows SmartScreen, graph strictness, and other unglamorous details.