← all repositories
RunMaestro/Maestro

Desktop air traffic control for your coding agents

Maestro exists to replace a cluttered terminal full of AI agents with a keyboard-driven, git-aware command center that can run automated playbooks for hours unattended.

3.3k stars TypeScript AgentsCoding Assistants
Maestro
Not currently ranked — collecting fresh signals.
star history

What it does Maestro is a cross-platform desktop application that wraps supported AI coding agents—Claude Code, OpenAI Codex, OpenCode, and Factory Droid—into a single keyboard-centric interface. It acts as a pass-through to your existing provider setups, so MCP tools, skills, and authentication remain identical, but adds fleet management: parallel sessions, git-aware worktrees, and automated execution of markdown task lists. The interface is explicitly built for developers who live on the keyboard and need to juggle long-running, unattended agent sessions across several projects at once.

The interesting bit The Auto Run feature treats markdown checklists as batch jobs, spawning a fresh, isolated AI session for each task to keep context from rotting; the author claims a personal record of nearly 24 hours of continuous unattended runtime. For parallel development, git worktrees let sub-agents operate on isolated branches while you continue working in the main repository, then surface pull requests without manual conflict resolution. A group-chat mode adds a moderator AI that routes questions between agents and synthesizes cross-project answers.

Key highlights

  • Keyboard-first design with Cmd+K quick actions, mastery tracking, and dual-mode terminals that switch between AI conversation and shell commands via hotkey
  • Git-aware parallel development: automatic repo detection, diff viewer, branch-based worktrees for agent isolation, and one-click PR creation
  • Auto Run and Playbooks: a filesystem-based runner that queues and batch-processes markdown tasks through clean agent sessions, plus a headless CLI for cron and CI/CD
  • Mobile remote monitoring via a built-in web server with QR-code access, plus message queueing so no prompt is lost while an agent is busy
  • Cost tracking per session, usage analytics with CSV export, and a document graph that visualizes internal wiki-links as an interactive knowledge map

Caveats

  • Maestro does not bundle AI backends; you must already have Claude Code, Codex, or OpenCode installed and authenticated separately
  • Support for additional agents such as Gemini CLI and Qwen3 Coder is explicitly listed as future work, pending community demand

Verdict If you are already running multiple AI coding sessions in separate terminal tabs and losing track of context, Maestro is purpose-built to replace that chaos with a git-aware, keyboard-driven control plane. If you only occasionally chat with a single coding agent, it is probably overkill.

Frequently asked

What is RunMaestro/Maestro?
Maestro exists to replace a cluttered terminal full of AI agents with a keyboard-driven, git-aware command center that can run automated playbooks for hours unattended.
Is Maestro open source?
Yes — RunMaestro/Maestro is open source, released under the AGPL-3.0 license.
What language is Maestro written in?
RunMaestro/Maestro is primarily written in TypeScript.
How popular is Maestro?
RunMaestro/Maestro has 3.3k stars on GitHub.
Where can I find Maestro?
RunMaestro/Maestro is on GitHub at https://github.com/RunMaestro/Maestro.

heatdrop uses Google Analytics to see which pages get read — nothing else. Your call. How we handle data.