System Monitoring

Built-in monitoring capabilities in LeafMesh.

Built-in Analytics

Usage Analytics

Aggregated analytics, cache metrics, and per-session event streams are available in Studio's Dashboard and session inspector — operator UI, not part of the customer Python or HTTP surface.

Returns aggregated statistics:

Metric	Description
`total_events`	Total events processed
`events_by_type`	Event counts grouped by type
`processing_rates`	Events per second
`error_rate`	Ratio of error events to total events

LLM Cache Statistics

Aggregated analytics, cache metrics, and per-session event streams are available in Studio's Dashboard and session inspector — operator UI, not part of the customer Python or HTTP surface.

Metric	Description
`hit_rate`	Cache hit ratio
`total_requests`	Total LLM requests
`cache_hits`	Requests served from cache
`estimated_savings`	Estimated cost savings from caching

Agent Statistics

agent_stats = leafmesh.get_agent_stats()

Per-agent metrics:

Metric	Description
`execution_count`	Total executions
`error_count`	Total errors
`error_rate`	Error ratio
`avg_response_ms`	Average response time
`current_status`	Current health status

Health Check Endpoint

The API server provides a liveness endpoint with overall agent health:

GET http://localhost:18820/health

{
  "status": "healthy",
  "agents": {
    "total_registered": 8,
    "healthy": 7,
    "degraded": 1,
    "failed": 0
  }
}

Periodic Monitoring Script

async def monitor_system(leafmesh):
    """Run as a scheduled agent or external script"""
    analytics = Studio's Dashboard (HTTP)
    agent_stats = leafmesh.get_agent_stats()

    # Check error rates
    for agent, stats in agent_stats.items():
        if stats.get("error_rate", 0) > 0.1:
            print(f"WARNING: {agent} error rate: {stats['error_rate']}")

Next Steps

Distributed Tracing — What you can trace
Performance Metrics — Detailed metrics
Alerting — Alert configuration

LeafMesh — Built-in system monitoring