Monitoring and Observability

Metrics

• Request latency (p50, p95, p99)
• Agent response time
• Job queue depth
• Generation success/failure rates
• System resource utilization

Logging

• Structured JSON logging
• Correlation IDs for request tracing
• Log aggregation and search
• Alert rules for critical errors

Health Checks

• Liveness probes for container health
• Readiness probes for traffic routing
• Deep health checks including dependencies