Every laptop has its own CLAUDE.md.None of them is yours.

Write your team's playbook once: instructions, commit and PR style, review checklist, shared prompts and guardrails. MControl gives it to every run, writes it into each agent's global file and shows who is on which version.

Try it: type a task and press Return.

Free for macOS 26 or later

Mission Control. Your team and its agents in one window.

Who works on what, which agents are running, the pull requests waiting for you and a standup written by local AI. Press ⌘K to start anything.

Mission Control

Team

  • Maya

    Available

    Rate-limit headers for the public API

    acme/api · agent/rate-limit-headers

    Codex12m

  • Jonas

    Deep Work

    Banner when the app is offline

    acme/ios · agent/offline-banner

    Grok Build3m

  • Lea

    Available

    Filters for search results

    acme/web · lea/search-filters

    Claude Code21min terminal

  • Diego

    In a Meeting

    Shares only their status

Running agents

Standup

Yesterday the team merged 2 pull requests, passkey sign-in among them. Right now 4 agents are running: payment retries, rate-limit headers, the offline banner and search filters. 2 pull requests wait for your review, and #211 already has green CI.

Local AI · this morning

Your queue

Activity

  • Jonas started Grok Build on “Banner when the app is offline”3m ago
  • You started Claude Code on “Retry failed payments in checkout”6m ago
  • Maya started Codex on “Rate-limit headers for the public API”12m ago
  • Maya approved #21018m ago
  • Diego opened pull request #21125m ago
4 running5 online2 reviews

Try it: open a run, switch screens, press ⌘K.

Every agent follows different rules.Now they follow yours.

Write your team's playbook once: system instructions, commit and PR style, a review checklist, shared prompts and guardrails. Every run in MControl gets it automatically. MControl also writes it into ~/.claude/CLAUDE.md and ~/.codex/AGENTS.md, so agents in the terminal follow the same rules.

Playbook

v14

System instructions

  • Read the code around a change before you edit it.
  • Keep changes small and focused on the task.

Output standards

  • Commits follow Conventional Commits.
  • Every change that can break comes with a test.

Review checklist

  • Does the change do what the task asks, and nothing else?
  • Are there tests for new behavior?

Shared prompts

  • /fix-tests Fix failing tests
  • /review Review my changes
  • /explain Explain this code

Guardrails

Highest mode
Forbidden commands
  • rm -rf
  • git push --force

Always in a worktree

Try it: change the highest mode or forbid another command.

Every run in MControl

  • claude --append-system-prompt
  • codex -c developer_instructions=…
  • grok --rules

Agents in the terminal

~/.claude/CLAUDE.md

<!-- MControl playbook v14 -->

- Commits follow Conventional Commits.

- Highest mode: Full Auto

- Forbidden commands: rm -rf, git push --force

~/.codex/AGENTS.md

<!-- MControl playbook v14 -->

- Commits follow Conventional Commits.

- Highest mode: Full Auto

- Forbidden commands: rm -rf, git push --force

Shared prompts as slash commands:/fix-tests/review/explain

Who's on which version

  • Youv14
  • Mayav14
  • Jonasv14
  • Leav131 behind
  • Diegov14

Free on your Mac. Pro for your team.

Run agents, keep a playbook and work through your board for free. With Pro, your whole team shares one playbook, one board and live status.

Free

On your Mac

  • Claude Code, Codex and Grok Build, side by side
  • Playbook and alignment for your agents
  • Task board, pull requests and Git
  • Mission Control and the menu bar

MControl Pro

For your team

  • Create or join a team
  • One playbook, one board, live status for everyone
  • Lead view with every agent of the team
  • Live relay and appearance

$12.99per Mac a month, cancel anytime

Every Mac on a team needs its own subscription.

See the team plan Just for you? Personal Pro is $9.99 a month, for private use. ›

No server. End-to-end encrypted.

Create a team with a name and invite people with a link. Tasks, playbook and live status are locked on your Macs before they sync through MControl Cloud, so only your team can read them. For instant updates inside your network, you can still host the optional relay yourself.

  • Only your team can read it

    Team data is encrypted on your Mac with a key only your team has. Not even we can read it. Sign-in is anonymous: no email, no names.

  • Your own accounts

    Agents run with your own Anthropic, OpenAI and xAI accounts. Your code never reaches our servers.

  • Local AI

    The standup is written by local AI, right on your Mac.

Version 3.5.0, 13.9 MB · What's new?

Downloaded more than 3,000 times.

What you need

macOS 26, git and at least one agent (claude, codex or grok), installed and signed in. For pull requests, the GitHub CLI (gh).

Help with the first launch

Open the downloaded file and drag MControl into Applications. Then start it from there. MControl is notarized by Apple and opens right away.

Also in your browser.

Mission Control, cloud runs with Claude Code, tasks, pull requests, the playbook and your team at app.mcontrol.tech. Look around in the Preview with sample data. No sign-in needed.