Codeman 🤖 Agent Open source

What it does

Codeman is a self-hosted dashboard for running terminal AI coding agents from any browser, including a phone. It spawns Claude Code, OpenCode, Codex, Antigravity, Gemini, Pi, Grok, DeepSeek Harness or OMP inside persistent tmux sessions and streams the real terminal to a web UI, so a session survives a dropped connection, a closed laptop or a reboot. Its main purpose is unattended and remote work: it re-prompts agents when they go idle, resumes them when a subscription usage limit resets, runs scheduled jobs, and shows every background agent and subagent working in real time. It runs entirely on your own machine and binds to loopback by default.

What is inside

• One dashboard for nine coding CLIs, run locally, in Docker, or over SSH, plus plain shell tabs and embedded dashboard URLs as web tabs
• A touch-optimised mobile UI with local echo, swipe navigation, push notifications and QR-code login instead of typing a password
• An idle-detection and respawn controller for multi-hour unattended runs, plus auto-resume on limit reset
• Live floating windows for subagents and teammates with real-time transcripts
• Multi-user mode with named, scrypt-hashed logins and an admin audit log; optional Docker-isolated and remote SSH sessions

Works with

The agents it launches: Claude Code, OpenCode, Codex, Google Antigravity, Gemini CLI and several others. It needs at least one of them installed. Requires Node.js 22+ and tmux (Windows via WSL).

How to install

curl -fsSL https://getcodeman.com/install | bash

codeman web
# Open http://localhost:3000 and start your first session

The installer prompts before every system change; an npm i -g aicodeman install is also available.

Maintenance and safety

MIT-licensed, actively developed, self-hosted with no telemetry. The install is a curl-to-shell script from the project's own domain, so read it first. Loopback binding is the default; binding to the network or opening a Cloudflare tunnel requires setting a password, and the docs are explicit about the risks. Its QR login design cites and addresses known QR-auth weaknesses.

Who should use it

Developers who run long or parallel agent jobs and want to monitor or drive them from a phone or another machine, and anyone who wants agents to keep working after they step away.