Мониторинг и наблюдаемость: Prometheus, Grafana, OpenTelemetry

observability

Monitoring expert is a Claude Code skill that sets up observability infrastructure: structured logging pipelines, Prometheus/Grafana stacks, alerting rules, distributed tracing via OpenTelemetry, and load testing with k6 and Artillery. Reach for it when a service is misbehaving in production and you have no visibility into why, when a new microservice needs metrics and dashboards before go-live, when alerts are firing too often or not at all, or when you need to find the exact bottleneck under sustained load. The skill produces working code: prom-client counters and histograms for Node.js, OpenTelemetry spans exported to Jaeger, PromQL-based Prometheus alert rules, and k6 scripts with p95 latency and error-rate thresholds. Dashboards are designed around RED (Rate/Errors/Duration) and USE (Utilization/Saturation/Errors) methodologies. Scope also covers CPU/memory profiling and infrastructure capacity planning. The direction is instrumenting and observing a running system — not pre-deploy debugging (use debugging-wizard for that). Best suited to Node.js stacks; examples for other runtimes are not documented in the source. MIT license, author Jeffallan, version 1.1.0.