Open source alternative to Conductor
Last verified
Orchestrate multiple AI coding agents in parallel from a cross-platform desktop app, with Auto Run, playbooks, git worktrees, and mobile remote control.
Maestro gives you a single window to manage all your AI agents, switch between conversations instantly, and queue messages while an agent is busy. Auto Run handles tasks you can define in a spec, so you launch it, let it run overnight, and review the results in the morning.
If you run a single AI session at a time and have no plans to parallelise, the overhead of a fleet manager is not worth it over a single terminal tab.
Worktree sub-agents each operate in an isolated branch directory, so three agents can be working on three separate features without interference. Each agent can open a pull request from inside Maestro when its branch is ready. Maestro Cue can trigger an agent when another agent's branch lands a commit.
If your team uses a shared remote environment rather than local git worktrees, the worktree integration adds no benefit without local repo access.
maestro-cli runs playbooks from cron jobs or CI/CD pipelines with JSONL output for scripting. SSH Remote Execution lets you point Maestro at build machines or GPU boxes. Maestro Cue fires agents on GitHub activity, turning agent invocations into first-class pipeline steps.
If you need agents to run purely server-side without any desktop dependency, Maestro's local architecture is a constraint. SSH Remote Execution may cover the case but requires Maestro running on a client machine.
Auto Run's Goal-Driven mode pursues a free-text objective through multiple tasks without a predefined checklist. Agent Resilience handles provider errors and quota resets automatically, so a 429 or 529 from your provider does not abort the session. The leaderboard shows community members logging sessions of 24 hours or more.
If your tasks are well-defined, short, and synchronous, the overhead of Auto Run setup is not worth it over a single interactive session.
Running multiple AI coding sessions in parallel is friction-heavy without tooling designed for it. Each agent lives in its own terminal tab with its own context, history, and state. Switching between them breaks flow, messages pile up while you are in another tab, and there is no way to see the overall status of your fleet at a glance.
The coordination problem compounds when tasks span branches or repositories. Git worktrees let you run parallel work on isolated branches, but nothing ties your agents to those branches or lets you launch a sub-agent per branch from one interface. Teams running three to five parallel AI sessions routinely lose track of which agent did what, which session has stalled waiting for input, and what each one's current task is.
Runs agents autonomously on markdown checklists (Spec-Driven) or free-text objectives (Goal-Driven). Each task gets its own AI session for clean context, so long multi-step work does not bleed conversation history. Auto Run sessions can run for hours unattended; the README documents a single session of nearly 24 hours. Playbooks make workflows repeatable and runnable from cron jobs or CI/CD pipelines via maestro-cli.
Runs Claude Code, OpenAI Codex, OpenCode, Factory Droid, Copilot-CLI, Qwen3 Coder, and Oh My Pi simultaneously with no defined limit. Each agent has an isolated workspace, conversation history, task memory, and context. Maestro is a pass-through to your provider configuration: MCP tools, permissions, and authentication carry over without reconfiguration.
Creates worktree sub-agents from the git branch menu, each in an isolated directory. Agents can run Auto Run sessions on their own branches independently while you work interactively in the main repository. Pull requests can be opened from within the app, and the diff viewer shows branch changes alongside active conversations.
A built-in web server exposes a control UI accessible by QR code. Works on the local network or over a Cloudflare tunnel for remote access. You can monitor agent progress, review output, and send messages to your fleet from a phone without being at the machine.
Triggers agent prompts on file changes, timers, GitHub activity, or another agent finishing its task. Chains those triggers into self-running pipelines. Off by default and enabled in Settings. Covers the gap between manually launching agents and fully automated CI: a file lands, an agent picks it up.
maestro-cli lists agents and groups, runs playbooks headlessly, and outputs JSONL for scripting. SSH Remote Execution lets you orchestrate agents on remote hosts, build machines, or GPU boxes without opening a second desktop session.
Maestro and Conductor serve the same core use case: a dedicated interface for running multiple AI coding agents in parallel, with session management and monitoring above the raw CLI. Conductor is a proprietary commercial tool; Maestro is an AGPL-3.0 open source alternative you install locally.
Conductor is shutting down on November 20, 2026. It is not a viable option for new setups.
| Feature | Maestro | Conductor |
|---|---|---|
| License | AGPL-3.0 | Proprietary |
| Deployment | Local desktop binary | Commercial service |
| Supported agents | Claude Code, Codex, OpenCode, and more | Claude Code |
| Auto Run and playbooks | Yes | Yes |
| Git worktree support | Yes | Yes |
| Mobile remote control | Yes | No |
| Event-driven automation | Yes (Maestro Cue) | No |
| SSH remote execution | Yes | No |
| Headless CLI | Yes (maestro-cli) | No |
| Cost | Free (AI provider costs apply) | Paid subscription |
Maestro covers Conductor's headline features and extends them: multi-provider support across seven agents (Conductor supported Claude Code only), mobile remote control, event-driven automation through Maestro Cue, SSH remote execution for remote hosts, and a full headless CLI for scripting and CI/CD pipelines. For anyone evaluating tools for a new setup, Maestro is the right choice.
The AGPL-3.0 license covers personal and commercial team use without restriction. If you plan to offer a modified version as a managed service to external users, AGPL-3.0 requires publishing the source changes.
Download the latest release for your platform from GitHub Releases: https://github.com/RunMaestro/Maestro/releases/latestMaestro currently supports Claude Code, OpenAI Codex, OpenCode, Factory Droid, Copilot-CLI (beta), Qwen3 Coder (beta), and Oh My Pi (beta). It is a pass-through to the agent's existing CLI, so you need each agent installed and authenticated before Maestro can connect to it. The README notes plans for Gemini CLI support based on community demand.
Yes. Maestro is AGPL-3.0 licensed and free to download and use. You still pay for the AI providers you connect to (Claude Code, OpenAI Codex, etc.) at their own rates. Maestro itself adds no subscription or per-agent fee. The only restriction under AGPL-3.0 is that if you distribute a modified version as a hosted service to others, you must publish the source changes.
Download the latest release for your platform from the GitHub Releases page at https://github.com/RunMaestro/Maestro/releases. Official releases cover macOS, Windows, and Linux. Building from source is documented in the README for contributors; the binary is the recommended install for end users.
Auto Run executes a task spec through your agents: Spec-Driven mode processes a markdown checklist one item at a time, each in a fresh AI session; Goal-Driven mode pursues a free-text objective autonomously. A playbook is a saved, named Auto Run configuration you can run repeatedly, trigger from maestro-cli, or schedule in a cron job. The two features work together: you create playbooks for recurring workflows and Auto Run executes them.
Yes, through SSH Remote Execution. You point Maestro at a remote host over SSH and it orchestrates agents on that machine, letting you use build servers, GPU boxes, or any specialized environment without leaving your local fleet view. Agents still need to be installed and authenticated on the remote host.
Self-hosted AI agent management with org charts and budgets
#1 in the Conductor guide
Drop-in Conductor replacement with 27 or more agents on macOS, Windows, and Linux under MIT
#2 in the Conductor guide
MIT-licensed control plane for 6 or more agents with mobile and web clients and a large community
#3 in the Conductor guide
Daemon-based architecture for agents running on a VM or Docker host with full mobile parity
Sponsors
Alternative to
Superhuman
AI email assistant that drafts replies and clears your inbox
Alternative to
Grok Bot
AI teammates you own: your keys, your model, your machine.
Live monitoring on LinkedIn, X, Reddit & more. Agentic social listening for startups.
The projects we add and every new guide. Free, unsubscribe any time.
Claude Code to 
OpenCode
Jira to 
Windshift
Notion to 
AppFlowy
Slack to 
Mattermost
Linear to 
GitLab
Wispr Flow to 
Handy
Arc to 
Zen Browser
CapCut to 
Concat
Grok Bot to 
rakazo
Superhuman to 
Inbox Zero
Airtable to 
AITable
Muse to 
OpenClaw
Railway to 
Coolify
GitHub Copilot to 
Cline
Datadog to 
NetdataConfluence to 
Wiki.js
Microsoft Teams to 
Rocket.Chat
Firebase to 
Supabase
Pinecone to 
Qdrant
Zapier to 
n8n
1Password to 
KeePassXC
Devin AI to 
OpenHands
Asana to 
Leantime
Contentful to 
Strapi
