Observability Readiness

When to use

Invoke before promoting a service to production, or when an existing service is page-prone, slow to diagnose, or has alert fatigue. Use after the service is running locally; do not invoke for greenfield design (use system-design-from-prd instead).

Inputs

• Running service spec (endpoints, latency expectations, error budget).
• Target metrics backend (Prometheus assumed) and tracing backend (Tempo/Jaeger/OTel collector).
• Log aggregator (Loki/CloudWatch/etc.).

Output contract

Generated instrumentation MUST conform to observability-standards. That standard is authoritative for: