Maestro
Open-source desktop app for running several AI coding agents such as Claude Code or Codex in parallel and working through task lists automatically.
What it does
Maestro is a desktop app that brings coding agents you already have installed – Claude Code, Codex, OpenCode or Factory Droid – together in one place; the project lists Copilot-CLI, Qwen3 Coder and Oh My Pi as beta. Each agent gets its own workspace with history and separate context. With Auto Run, agents work through Markdown checklists item by item, and reusable playbooks can also be started from the command line via cron jobs or CI pipelines. There is also a Git view with diffs and worktrees for parallel branches, group chats between agents, execution on remote machines over SSH, and remote control from your phone through a built-in web server.
Who it suits
Development teams and freelancers who already use coding agents and want to run several tasks or projects at once without switching between many terminal windows.
Data protection for businesses
Maestro runs locally on your machine and needs no account. According to the project’s privacy page, the app itself does not transmit code, prompts, replies or API keys. Switched on by default are an install check-in (random ID, version, operating system) and hashed automation counters sent to runmaestro.ai (hosted on Cloudflare), plus crash reports sent to Sentry; all of this can be switched off in the settings. The privacy page was last reviewed on 22 September 2026 against version 0.18 and names no responsible entity, only Discord and GitHub as contacts. The actual AI processing is done by the connected agent provider; server location, training use and data processing agreement depend on that provider’s terms and plans. The project does not mention its own data processing agreement or business plans.
Limits
Maestro does not work without a subscription or API access from an agent provider. There is no named company and no legal notice behind the project, but a developer in Austin with a small core team and a community; help is available via Discord and GitHub. Some extra features are only available in pre-release versions. Version 1.0 was released in October 2026.