Incident Runbooks AI Skill Guide

Overview & Engine Architecture

An incident runbook is a short, actionable guide an oncall engineer follows under pressure. Good runbooks start from observable symptoms, list safe diagnostics, then mitigation and escalation - not essays. Agents write runbooks that link to dashboards and feature flags, avoid secret values, and keep steps numbered and reversible when possible.

Alert / user report
   -> Symptom match in runbook index
       -> Severity + impact
           -> Diagnose (read-only first)
               -> Mitigate / roll back
                   -> Escalate + handoff notes
                       -> Post-incident follow-ups

When to use this skill