Agent Teams Setup Guide
Quick setup for parallel multi-agent development with SpecWeave.
Prerequisites
- Claude Code CLI installed
- SpecWeave initialized in your project (
specweave init) - Terminal: tmux (recommended) or iTerm2 (macOS)
Step 1: Install Terminal Multiplexer
tmux (Recommended — Cross-Platform)
# macOS
brew install tmux
# Ubuntu/Debian
sudo apt install tmux
# Windows (WSL)
sudo apt install tmux
iTerm2 (macOS Alternative)
- Install from iterm2.com
- Enable Shell Integration: iTerm2 → Install Shell Integration
- Enable Python API: Settings → General → Magic → Enable Python API
Step 2: Enable Agent Teams
Add to your .claude/settings.json:
{
"env": {
"CLAUDE_CODE_EXPERIMENTAL_AGENT_TEAMS": "1"
}
}
Restart Claude Code after changing settings.
Step 3: Run Your First Team
Quick Start with Presets
/sw:team-build --preset full-stack "Build user authentication"
Custom Orchestration
/sw:team-lead "Build checkout with Stripe payments, React frontend, and Node.js backend"
Step 4: Monitor Progress
/sw:team-status # One-time check
/sw:team-status --watch # Auto-refresh
Step 5: Merge Results
/sw:team-merge # Merge all agent work
Available Presets
| Preset | Agents | Use When |
|---|---|---|
full-stack | 3 (shared → backend + frontend) | Building features across stack |
review | 3 (security + quality + docs) | Code review before release |
testing | 3 (unit + e2e + coverage) | Comprehensive test generation |
tdd | 3 (red → green → refactor) | Test-driven development |
migration | 3 (schema → backend + frontend) | Database or API migrations |
Terminal Navigation
| Mode | How to Switch Panes |
|---|---|
| tmux | Ctrl+B then arrow keys |
| iTerm2 | Click on pane |
| In-process | Shift+Up / Shift+Down |
Troubleshooting
| Issue | Fix |
|---|---|
| Agents not spawning | Verify CLAUDE_CODE_EXPERIMENTAL_AGENT_TEAMS=1 is set. Restart Claude Code. |
| tmux panes freeze | tmux kill-server and retry. Use in-process mode as fallback. |
| File conflicts between agents | Check file ownership with --dry-run flag before launching. |
| Token costs too high | Use smaller presets or --max-agents 2. |
Further Reading
- Agent Teams & Swarms Guide — Architecture and concepts
/sw:team-lead— Custom team formation/sw:team-build— Preset-based team launch