CLI Agent Orchestrator
CLI Agent Orchestrator (CAO)
A CLI that coordinates several coding agents: a supervisor delegates tasks to specialists in parallel or in sequence across isolated tmux sessions
High risk
We rate an entry high when the tool writes to external systems, handles money, production databases or secrets, or runs arbitrary commands. The CLI installs it only with your consent.
Why this level
- Launches coding agents that run commands and change project files
- Starts a local server and coordinates several processes in tmux
Install
In your terminal, with SkillFoxx CLI
npx skillfoxx add cli/cli-agent-orchestrator-caoDetects the agents on your machine, checks the risk and pins the version.
Other ways to install
Install the tool
pipx install cli-agent-orchestratorWithout pipx, pip install cli-agent-orchestrator works too.
Install CAO as a uv tool: uv tool install git+https://github.com/awslabs/cli-agent-orchestrator.git@main --upgrade. Install the supervisor profile with cao install code_supervisor, run cao-server in one terminal and cao launch --agents code_supervisor in another from your project folder.
Other ways from the author
uv tool install git+https://github.com/awslabs/cli-agent-orchestrator.git@main --upgrade
cao --helpInstalls the main branch as a uv tool. Tagged releases are on PyPI as cli-agent-orchestrator.
This is third-party code. Review the repository files before installing.
What it does
CAO runs a local cao-server, launches the chosen coding CLIs in isolated tmux sessions and gives a supervisor tools to coordinate workers. The agents stay full processes with their own authentication and capabilities; CAO only splits the work between them. It supports several providers: Claude Code, Codex, Cursor CLI, GitHub Copilot CLI, OpenCode, Kiro and others. It ships a localhost web UI, an operations MCP server, scheduled flows, multi-step workflows, shared memory and outcome-based learning. You drive it through cao commands: install a supervisor profile, launch and shut down sessions.
Who it is for. For developers and teams who want to run several coding agents in parallel under a shared supervisor.
Good fit when
- You want to parallelize work across several coding agents
- You need a supervisor that hands subtasks to specialists and merges the result
- You want to combine agents from different providers in one flow
Not a fit when
- A single agent without orchestration is enough
- You have no tmux, uv or an installed provider CLI
Example request
Launch a supervisor and split the work: one agent writes tests, another reviews the codeLimitations
It needs Python 3.10+, tmux 3.3+, uv and at least one supported provider CLI authenticated in advance. Providers use their own accounts and keys, and some may require access that is not directly available from Russia. Installing the main branch is a uv tool install; tagged releases are on PyPI.
How to disable. Stop the sessions with cao shutdown --all, then remove the tool: uv tool uninstall cli-agent-orchestrator.
Security check
- Launches coding agents that run commands and change project files
- Starts a local server and coordinates several processes in tmux
README in short
The README explains that CAO coordinates several coding CLIs: a supervisor hands work to specialists in parallel or in sequence. It needs Python 3.10+, tmux 3.3+, uv and at least one authenticated provider CLI. Installation is a uv tool install from the main branch or from PyPI, then you install the code_supervisor profile, start cao-server and launch the supervisor with cao launch. You can watch the work in the terminal, in a localhost web UI or over tmux. The docs cover flows, workflows, skills, memory, tool restrictions and a Kubernetes deployment. Apache-2.0 licensed.
FAQ
Which agents does it work with?
Coding CLIs: Claude Code, Codex, Cursor CLI, GitHub Copilot CLI, OpenCode, Kiro and several others. Each runs under its own account.
Does it need a cloud service?
No, cao-server runs locally and there is a web UI on localhost.
Related
Open-source personal AI assistant on your own machine: answers in Telegram, Slack, Discord and WhatsApp, extended with skills and plugins
A self-improving agent from Nous Research with a TUI, messaging gateway, cron jobs and skills it writes itself
An open source coding agent for the terminal and desktop with build and plan modes
Open prompt library with a Claude Code plugin, MCP server and CLI: search, fetch and improve prompts and skills from an agent