Multi-agent team launcher for Claude Code. Coordinate AI agents to work together on your projects.
Claude Team lets you run multiple Claude agents as a coordinated team:
- CONDUCTOR: The manager - plans tasks, delegates work, reviews output
- ENGINEER: The builder - writes code, implements features
- TESTER: The validator - tests, finds bugs, ensures quality
Each agent has its own role, personality, and responsibilities. They communicate through a shared KANBAN board.
Claude Team follows a minimal scrum methodology where agents collaborate through structured workflows:
The system automatically extends context across agents, allowing them to share knowledge and maintain consistency:
- Node.js 20+
- Claude Code CLI
- tmux
- ttyd (for web dashboard terminal streaming)
brew install node tmux ttyd
npm install -g @anthropic-ai/claude-code# Install
npm install && npm run build && npm link
# Start your AI team
ct --start
# Launch single agent
ct CONDUCTOR
ct ENGINEER
ct TESTER
# Resume existing session
ct --start --resumeReal-time interface to chat with your AI team and monitor progress.
# Terminal 1: Start team
ct --start
# Terminal 2: Launch dashboard
ct --dashboardOpen http://localhost:3000 in your browser.
| Command | Description |
|---|---|
ct --start |
Launch full team in tmux |
ct --start -w duo |
Launch duo workflow (2 agents) |
ct CONDUCTOR |
Launch single agent |
ct --status |
Show session status |
ct --kill |
Stop the team |
ct --list |
List tmux sessions |
ct --list-workflows |
List available workflows |
ct-send -r ENGINEER "message" |
Send message to agent in current session |
ct-send -s session_name -r ENGINEER "msg" |
Send message to agent in specific session |
ct-capture |
Capture session transcript |
ct-reminder |
Set periodic reminders |
basic (default): 3-agent workflow
- CONDUCTOR: Plans and coordinates
- ENGINEER: Implements solutions
- TESTER: Tests and validates
duo: 2-agent workflow
- LEAD: Plans and reviews
- DEV: Implements features
ct --start -w duoMIT