observability

@pmq9

Use this skill for any work involving logs, metrics, traces, alerting, dashboards, SLOs, or monitoring — including instrumenting a service with OpenTelemetry, Prometheus, structured logging, or distributed tracing; writing or critiquing alerts; designing dashboards; defining SLIs/SLOs/error budgets; investigating an incident with telemetry; setting up Grafana / Loki / Tempo / Jaeger / Datadog / Honeycomb / New Relic; debugging a high-cardinality metrics blow-up; or deciding what to log and what not to. Trigger on "instrument", "trace", "metrics", "logs", "alert", "PagerDuty", "Prometheus", "Grafana", "OTel", "SLO", "p99 latency", "what should we monitor". Also trigger when a user describes a production problem and is about to debug without telemetry — push them to look at the data first.