prd-v08-runbook-creation_skill

Overview

This skill creates concise, actionable runbooks for incident response, deployments, maintenance, recovery, and escalation during PRD v0.8 Deployment & Ops. It outputs RUN- entries with clear scope, prerequisites, step-by-step procedures, verifications, escalation paths, and post-incident actions. The runbooks are written for operators with minimal context to ensure fast, reliable responses.

How this skill works

On request, the skill identifies critical scenarios (MON-, DEP-, RISK- entries) and maps each scenario to a single RUN- runbook. It generates a template-based runbook that includes scope, prerequisites, numbered procedures with commands, verification checks, escalation instructions, and post-incident tasks. Each output is traceable with linked IDs and a last-tested date to support maintenance and drills.

When to use it

• When an alert fires and you need a clear step-by-step response
• Before a scheduled production deployment or hotfix
• For routine maintenance windows or scheduled backups
• When defining escalation paths for on-call teams
• During release planning to ensure MON-/DEP-/RISK- coverage

Best practices

• Create one runbook per distinct scenario with explicit scope and exclusions
• Include precise commands, links, and contact info; assume minimal operator context
• Add verification checks after each major step to confirm success
• Define clear escalation conditions, contacts, and required handoff data
• Keep an owner and last-tested date; update runbooks after drills or incidents
• Split overly long procedures into focused runbooks to avoid operator confusion

Example use cases

• RUN-001 style incident: database connection pool exhaustion with SQL checks, kill steps, and escalation
• RUN-002 style deployment: pre-deploy announcement, checkpoint tagging, CI/CD deploy, smoke tests, and 15-minute monitoring
• Maintenance runbook: certificate renewal with prerequisites, step-by-step renewal, verification, and rollback steps
• Recovery runbook: data restore procedure with verification and post-restore validation
• Escalation runbook: when automated remedies fail, who to contact and what information to provide