Observability¶

The repository includes a full observability stack for local cluster use.

Components¶

• Loki for logs
• Tempo for traces
• Prometheus for metrics
• Grafana for dashboards
• Promtail for log shipping

What to watch first¶

• backend error logs
• request latency spikes
• failed tool or model calls
• readiness probe failures
• broken SSE streams

Common entry points¶

• Grafana in the Kind deployment
• structured backend logs from setup_logging()
• frontend warn and error logs via /api/logs