observability-engineeringlisted

# Observability Engineering Make systems answer operational questions with emitted evidence. Start with the decisions an operator must make, not with a vendor, dashboard, or “log everything” mandate. ## Load only what the task needs - Event schemas, severity, correlation, async boundaries, redaction, or cost controls → [signal-contract.md](references/signal-contract.md) - SLI/SLOs, error budgets, alerts, dashboards, runbooks, probes, or smoke evidence → [reliability-and-proof.md](references/reliability-and-proof.md) - Standards, official guidance, and freshness checks → [source-ledger.md](references/source-ledger.md) Telemetry SDKs, semantic conventions, runtime hooks, exporter configuration, and vendor APIs are volatile. Recheck the installed versions and current official documentation before naming an API, configuration field, limit, or command. Project scripts and the installed stack win. ## Scope and ownership Use this skill for cross-stack signal contracts, observable user journeys, correlation across components, reliability measurement, diagnostic telemetry safety/cost, and proof that signals actually travel from application boundary to the configured sink. Do not use it to choose a monitoring vendor or reproduce a provider catalog. It does not own: - browser performance diagnosis or Core Web Vitals interpretation → `web-perf`; - Next.js or Astro runtime instrumentation APIs → `nextjs-development` or `astro-development`; - Cloudflare Workers/runtime configuratio