buster7 Documentation
How to use a lightweight agent development environment for running agents side by side
buster7 is a lightweight agent development environment for running several Claude Code or Codex instances side by side. It holds no model of its own. It runs the agent CLIs you already have, and stays a mother ship where a human watches them and gives the orders.
This site is for people who are about to use buster7.
Where to start
What buster7 is
What the tool does, and what it does not
Get it running
What you need, starting the server and the app, connecting
A map of the screen
Board, tabs, terminal, chat, files
Settings
Connection, appearance, notifications
CLI (b7)
Let an AI agent drive buster7
MCP integration
Register it as a tool for your agents
About this documentation
- Screens and operations are covered in Usage, and every setting is listed in Settings
- To let an agent drive buster7 itself, see CLI (b7) and MCP integration
- When something does not work, look at Troubleshooting
In development
A free build for macOS is available as a zip. It is still in development, and some problems remain. Get it running covers the download, the first launch, and connecting.