Agent Orchestrator is an open-source desktop application for running whole fleets of AI coding agents in parallel and supervising them from a single board. Published on GitHub under the Apache-2.0 license, it has gathered more than 13,400 stars and was in active development in 2026. Instead of replacing the coding agent you already trust, it sits above 36 of them โ including Claude Code, Codex, Cursor, OpenCode, Aider and Goose โ and coordinates their work end to end.
You describe the outcome you want, and a main orchestrator agent plans the work, spawns worker agents into isolated Git worktrees, and keeps them moving without file conflicts. Every session appears on one board with its agent, branch and pull-request state. Orchestrator then watches each PR it opens: failed CI checks, review comments and merge conflicts are routed back to the session that owns the branch, with the context needed to fix them, until the PR is approved. A mobile companion app pairs over LAN or Tailscale so you can check on the fleet and respond when an agent needs you.
The local app is free and open source with unlimited local agents and no account required. Optional Orchestrator Cloud plans start at $20/month and run agent sessions in hosted sandboxes for teams that do not want execution on their own machines.
Agent Orchestrator is for developers and small teams who already use coding agents heavily and want to parallelize them safely โ shipping several features at once while keeping human review firmly in the loop.