Monitoring Observability

This is the skill you reach for when the question isn't "why did it break?" but "what should we measure, and what should wake someone up?" It walks you through framing the observability work you actually have,new service setup, stale dashboard audit, pipeline freshness checks, game crash visibility,then forces you to pick one primary mode (service reliability, telemetry foundation, data pipeline, live ops, or review audit) before dumping metrics everywhere. The core output is a smallest signal plan: what questions the dashboard must answer, what pages a human versus creates a ticket, who owns each alert family, and explicit handoffs to log-analysis for root cause, debugging for regressions, or performance-optimization for bottlenecks. It's opinionated about symptoms over causes and bounded cardinality, and it won't let you blur incident response with product analytics.

Install to Claude Code

npx -y skills add akillness/oh-my-skills --skill monitoring-observability --agent claude-code