Observability Engineering Skill

Source: Observability Engineering — Charity Majors, Liz Fong-Jones, George Miranda (O'Reilly, 2022)

This skill encodes the book's core wisdom so Claude Code can act as an observability-aware engineer: knowing what to look for (from the book) and how to look (via MCP tools connected by the user).

1. Core Philosophy: Observability vs Monitoring

The fundamental definition: Observability is a measure of how well you can understand and explain any state your system can get into, no matter how novel or bizarre — without needing to predict that state in advance.

Litmus test — you have observability if:

• You can answer open-ended questions about internal workings without hitting investigative dead ends
• You can understand what any particular user is experiencing at any given time
• You can see any cross-section of system performance — from aggregate to single request
• You can isolate any fault within minutes, no matter how deep in the stack
• Your debugging investigations often surprise you with novel findings (not just confirming suspicions)