Swarm Dashboard Skill
Real-time monitoring for AI agent swarms with a beautiful, auto-refreshing web UI.
Quick Start
python
from swarm_dashboard import launch_dashboard, update_agent_task_id
# Launch dashboard
url = launch_dashboard(
swarm_name="My Project",
swarm_dir="/workspace/my-project",
agents={
"agent-1": {"role": "Core Architect", "wave": 1, "mission": "Setup"},
"agent-2": {"role": "Backend Dev", "wave": 2, "mission": "Build API"},
}
)
# After launching each agent with Task tool, update task ID
update_agent_task_id(
swarm_dir="/workspace/my-project",
agent_id="agent-1",
task_id="abc123" # From Task tool response
)
Features
- •Real-time Updates - Auto-refreshes every 2 seconds
- •Smart Completion Detection - Auto-detects finished agents
- •Capybara-Inspired UI - Natural color palette with light/dark themes
- •Clickable Details - Live activity feed, tools used, files created
- •Zero Dependencies - Pure Python stdlib
CLI Usage
bash
# Start server swarm-dashboard serve --port 8080 --swarm-dir /workspace # Launch in background swarm-dashboard launch --name "My Swarm" --dir /workspace # Check status swarm-dashboard status --dir /workspace # Stop swarm-dashboard stop --dir /workspace
Configuration
Environment variables:
- •
SWARM_DIR- Directory with agent folders - •
TASK_DIR- Directory with task outputs - •
DASHBOARD_PORT- Server port (default 8080) - •
SWARM_NAME- Display name
API Endpoints
| Endpoint | Description |
|---|---|
GET / | Dashboard HTML |
GET /api/status | Swarm status JSON |
GET /api/agent/{id} | Agent details |
GET /health | Health check |