Runbook Refresher
Auto-generate runbooks from your actual infrastructure, not from memory
Generates and maintains operational runbooks by reading your actual infrastructure — Terraform configs, K8s manifests, monitoring setup, and incident history. Detects when existing runbooks reference decommissioned resources and flags them for update.
INGREDIENTS
PROMPT
Create a skill called "Runbook Refresher". Generate and maintain operational runbooks: Mode 1 — Generate: Read my infrastructure configs (Terraform, K8s manifests, docker-compose, monitoring configs) and generate runbooks for: - Service restart and scaling procedures - Common failure scenarios and recovery steps - Deployment and rollback procedures - Health check and diagnostic commands - Dependency failure handling Format each runbook with: Prerequisites, Steps (with exact commands), Verification (how to confirm the fix worked), and Escalation path. Mode 2 — Audit: Review existing runbooks against current infrastructure and flag: - References to decommissioned resources (old hostnames, IPs, services) - Commands that would fail against current configs - Missing runbooks for services that don't have any - Outdated screenshots or procedure descriptions
How It Works
Documentation rots because it's maintained separately from infrastructure.
This skill generates runbooks directly from your configs and code, so
they're accurate by construction. It also audits existing runbooks against
current infrastructure.
What You Get
- Auto-generated runbooks from Terraform, K8s, and monitoring configs
- Service-specific runbooks: how to restart, scale, debug, and recover
- Incident playbooks based on historical incident patterns
- Stale documentation detection (references to decommissioned resources, renamed services, etc.)
- Standard operating procedures for common tasks (deploy, rollback, scale, failover)
Setup Steps
- Point your Claw at your infrastructure code and existing docs
- Let it generate or audit runbooks
- Review and add human context (tribal knowledge, edge cases)
- Run the audit quarterly to catch documentation drift
Tips
- The best runbooks are tested by actually running through the steps
- Include "how to verify the fix worked" in every runbook
- Link runbooks to specific alerts so on-call engineers find them fast
- Ask your Claw to generate runbooks in your team's standard format
- Pair with Shift Kickoff to surface relevant runbooks during incidents