Cross-service trace correlation during investigation

[nomadicmehul]

Summary

When investigating an incident, the agent should follow distributed traces across services to identify where failures originate vs. where they manifest.

Why This Matters

A 5xx error on the API gateway might be caused by a database connection pool exhaustion 3 hops away. Without trace following, the agent only sees the symptom, not the cause.

Acceptance Criteria

• Given an error in service-A, fetch the trace ID from logs/observability
• Follow the trace across all services it touched
• Identify the first failure point in the trace (origin of the error)
• Include trace evidence in RCA: "Error originated in service-D (DB timeout), propagated through C → B → A"
• Works with Grafana (Tempo), Datadog APM, or Cloud Trace

Dependencies

• Requires Phase 1 observability integrations