Инструментирование кода для наблюдаемости в production

★ 8.7 · devops

observability-and-instrumentation is a Claude Code skill that instruments code so production behavior is visible and diagnosable through structured logging, metrics, distributed tracing, and alerting. It applies RED (Rate, Errors, Duration) for request-driven services and USE (Utilization, Saturation, Errors) for resources, enforces correlation ID propagation across every request boundary, and guards against PII leaking into telemetry pipelines. The process starts by defining the exact questions an on-call engineer will ask, then selects the right signal type — logs answer "why", metrics answer "how often", traces answer "where" — before writing any instrumentation, using OpenTelemetry as the vendor-neutral foundation. Intended for teams who want to diagnose production incidents through telemetry queries rather than log archaeology, and who treat observability as a first-class deliverable shipped alongside every feature.