Observability: Metrics, Traces, Logging, SLO Dashboards, Alerting

[IgnacioPro]

Production readiness requires Prometheus metrics, OTel traces, log correlation, sample control, and default Grafana dashboards with SLO-driven alerts (latency, error rates, agent health, AI cost).

Paths to start:

• Metrics: expose Prometheus endpoints in API/agent; see any metrics.go or injection in main.go
• Traces: distributed and sample logic in API, agent, key flows (internal/diagnostics/, circuit breaker layer ref: CLAUDE.md)
• Logging: add and correlate context IDs; see logrus config
• Dashboards/alerts: publish default Grafana JSON; SLO recipes for latency, errors, agent heartbeat, budget

Integrate with key flows (diagnostics, remediation, agent lifecycle).

References: internal/diagnostics/, CLAUDE.md, grafana dashboards, main.go.