multi-agent-manager（多 Agent 经理）

Continuous Monitoring

Watch agent status in real-time:

python3 {baseDir}/scripts/monitor.py --watch --interval 5

JSON Output

Get machine-readable output:

python3 {baseDir}/scripts/visualize.py --format json

Example JSON:

{
  "agents": [
    {
      "id": "main",
      "model": "anthropic/claude-sonnet-4-6",
      "workspace": "/Users/user/.openclaw/workspace",
      "sessions": 0
    }
  ],
  "total": 2
}

📖 Use Cases

1. Debugging Agent Issues

When an agent isn't responding:

# Check if agent is running
python3 {baseDir}/scripts/monitor.py

# View agent details
python3 {baseDir}/scripts/visualize.py

2. Managing Multiple Projects

Track agents across different projects:

# See all agents and their workspaces
python3 {baseDir}/scripts/visualize.py

3. Optimizing Agent Performance

Monitor which agents are most active:

# Watch real-time activity
python3 {baseDir}/scripts/monitor.py --watch

🎯 Common Scenarios

Scenario: "Which agents do I have?"

python3 {baseDir}/scripts/visualize.py

Scenario: "Is my agent still running?"

python3 {baseDir}/scripts/monitor.py

Scenario: "What's my agent doing right now?"

python3 {baseDir}/scripts/monitor.py --watch

💡 Tips

• Use --format json for integration with other tools
• Run monitor.py --watch in a separate terminal for continuous monitoring
• Check agent status before sending important tasks

🐛 Troubleshooting

Problem: "No agents found"

• Make sure OpenClaw Gateway is running: openclaw status
• Check your agent configuration

Problem: "Script not found"

• Ensure you're running from the correct directory
• Check that Python 3 is installed: python3 --version

📝 Changelog

v1.0.0 (2026-03-14)

• Initial release
• Agent visualization
• Status monitoring
• Basic communication tools
• Task flow tracking

🤝 Contributing

Found a bug or have a feature request?

• Open an issue on GitHub
• Join the OpenClaw Discord community

📄 License

MIT License - feel free to use and modify!