Application metrics

Intent

Expose metrics that make service health and performance measurable (request volume, errors, latency, and domain signals).

Use when

• Always: metrics are a prerequisite for operating distributed systems reliably.
• You need alerting, capacity planning, and debugging signals beyond logs.

Avoid / watch-outs

• Avoid high-cardinality labels (user IDs, raw URLs, unbounded IDs).
• Metrics should be stable contracts; changing names/labels breaks dashboards/alerts.

Skill mapping

• observability: define the field/label contract and implement RED + domain metrics.
• spec: document SLOs and how metrics reflect success/failure semantics.