awslabs/cli-agent-orchestrator

Multi-agent orchestration for AI coding CLIs — Claude Code, Kiro, Codex, and more, coordinated in isolated tmux sessions

CLI Agent Orchestrator (CAO)

What it is

  • A Python‑based tool that lets you run multiple AI‑coding command‑line interfaces (like Claude Code, GitHub Copilot CLI, etc.) side‑by‑side and coordinate them from a single “supervisor” process.
  • It launches a local server (cao‑server) and uses tmux sessions to keep each provider CLI isolated while preserving their native authentication.

Key capabilities

  • Orchestration – start, stop, and monitor many provider CLIs in parallel or in a defined sequence.
  • Control planes – interact via a web UI, a shell CLI, an MCP (management) server, or plug‑in extensions.
  • Workflows & flows – define multi‑step pipelines that chain agents together.
  • Skills & memory – reusable guidance packages (skills) and optional persistent memory that can turn past runs into lessons.
  • Plugins – support for agent‑ and event‑plugins (currently gated behind an env‑var).
  • Deployments – can be run locally or on Kubernetes/EKS clusters.

Typical workflow

  1. Install a provider CLI (e.g., Kiro, Claude Code) and authenticate it.
  2. Install a supervisor profile (cao install code_supervisor).
  3. Start the CAO server (cao‑server).
  4. Launch the supervisor with the desired agents (cao launch --agents code_supervisor).
  5. Watch the coordination in a terminal, the web UI (http://localhost:9889), or attach via tmux.
  6. Shut down sessions with cao shutdown.

Who it’s for

  • Developers who want to harness several AI coding assistants at once, without manually juggling separate terminals.
  • Teams that need reproducible, scripted AI‑assisted coding pipelines or want to run the orchestration in CI/CD or on a Kubernetes cluster.

Installation

  • Requires Python 3.10+, tmux 3.3+, and the uv package manager.
  • Install directly from the repo with uv tool install git+https://github.com/awslabs/cli-agent-orchestrator.git@main or from PyPI (pip install cli-agent-orchestrator).

Documentation & learning

License

  • Apache License 2.0.

All details are taken from the repository’s README.

Related

  • Project
    vicoa-ai/vicoa

    An open-source AI orchestrator that lets you run and manage a team of coding agents across desktop, mobile, and remote servers from a single command center.

  • Project
    getpaseo/paseo

    An open-source agentic development environment that provides a unified interface for multiple AI coding agents across desktop, mobile, and CLI.

  • Project
    OrchestratorInc/agent-orchestrator

    A local desktop workspace for planning, running, and supervising multiple AI coding agents on a single project through isolated workspaces and a live Kanban board.

  • Project
    stablyai/orca

    An AI orchestrator that lets developers run multiple CLI-based coding agents in parallel isolated worktrees to compare results and merge the best output.

  • Project
    code-yeongyu/oh-my-openagent

    An AI agent command-line tool that automates complex professional workflows like coding and deep research using multi-agent graphs and a persistent memory system.