Summary

Agent Orchestrator tackles the coordination complexity of managing fleets of parallel AI coding agents. It provides an autonomous system for task planning, agent spawning with isolated worktrees/branches, and automatic handling of CI failures and code review comments. This enables developers to focus on high-level decisions while the system manages iterative coding, aiming to boost productivity.

How It Works

Each spawned agent receives an isolated git worktree and feature branch. Agents like Claude Code, Codex, or Aider run within configurable runtimes (default: tmux). The system autonomously monitors CI status and reviewer feedback, routing issues back to the agent for resolution. Users are notified only when human judgment is necessary, streamlining the workflow.

Quick Start & Requirements

Install via git clone and bash scripts/setup.sh. Configure with ao init --auto in your project directory, then ao start. Spawn agents using ao spawn <project> [issue]. Prerequisites: Node.js 20+, Git 2.25+, tmux, gh CLI. Dashboard at http://localhost:3000; CLI status via ao status. See Setup Guide and CLAUDE.md for details.

Highlighted Details

• Agent-agnostic (Claude Code, Codex, Aider) and runtime-agnostic (tmux, Docker, k8s).
• Tracker-agnostic (GitHub, Linear) with autonomous CI/review handling.
• Extensible plugin architecture for custom components.
• Web dashboard and CLI for monitoring and control.

Maintenance & Community

Contributions are welcomed, especially via the plugin system for adding new agents, runtimes, or trackers. Specific community channels or maintainer details are not provided in the README.

Licensing & Compatibility

Released under the MIT License, permitting broad use, modification, and distribution, including for commercial purposes, without copyleft restrictions.

Limitations & Caveats

The README does not detail known bugs or alpha status. Default setup requires Node.js, Git, tmux, and gh CLI, which may pose setup or compatibility considerations.