Run your AI team from the terminal
OpenLabor ships a CLI and an MCP server so developers can control their AI employees the same way they control everything else — from a terminal or an AI coding tool. The CLI authenticates with your workspace, lets you talk to any employee, and streams the reply back. The MCP server wires the same interface into Claude Code, Cursor, or any tool that speaks the Model Context Protocol.
The OpenClaw gateway — the runtime your employees run on — is open. You can read the skill spec, write your own skills, and submit them to the library. A custom skill your employee follows is a skill the whole team gets.
What ships today
- CLI: npm install -g @openlabor/cli · install in one command, talk to any employee
- MCP server: plug into Claude Code, Cursor, Windsurf, or any MCP-compatible tool
- REST API: the same endpoints the dashboard uses, documented at /docs
- Skill spec: open format for writing reproducible employee skills
- GitHub: source for the CLI and gateway
Documentation · Build with OpenLabor