Notre avis
Crée et met à jour des runbooks opérationnels pour les procédures de déploiement, base de données, maintenance et réponse aux incidents.
Points forts
- Utilise un modèle structuré et une liste de contrôle qualité
- Inclut des niveaux de risque et des étapes de rollback
- Intègre des principes de documentation stricts
- Produit des commandes copiables directement
Limites
- Nécessite que les principes de documentation soient déjà définis
- Demande l'avis de l'utilisateur pour les nouvelles procédures
- Les procédures à haut risque exigent une approbation d'équipe
À utiliser pour documenter une procédure opérationnelle répétable avec des étapes claires et des scénarios de récupération.
À éviter pour des tâches ponctuelles ou lorsque le processus n'est pas encore implémenté dans le code.
Analyse de sécurité
SûrThis skill is purely for documentation purposes. It does not instruct any tool execution, system commands, or data handling. The content is about creating runbooks, which is a passive, informative task with no inherent security risk.
Aucun point d'attention détecté
Exemples
Create a runbook for deploying to production, including steps, rollback, and failure recovery.Document the database migration process with commands, expected outputs, and rollback plan.Write an incident response runbook for a service outage, including troubleshooting and escalation steps.name: document-runbook description: Document CI/CD, deployment, and operational procedures. Use when creating runbooks, documenting deployment processes, or writing operational guides.
Runbook Documentation Skill
Purpose
Create operational runbooks for deployment, database operations, incident response, and maintenance procedures. These are "How-To" documents in the Diátaxis framework, specifically for operations tasks.
Smart Interaction
ASK the User When:
- Creating new runbook: Confirm process name and risk level
- Deleting runbook: Always confirm before deletion
- High-risk procedures: Confirm rollback steps are adequate
PROCEED Autonomously When:
- Updating existing runbook: Add new steps, update commands
- Adding troubleshooting: Enhance with new failure scenarios
- Fixing commands: Correct outdated or broken commands
- Adding verification steps: Improve procedure completeness
Documentation Principles (CRITICAL)
Before writing ANY documentation, review ../DOCUMENTATION_PRINCIPLES.md for:
- Ground Truth Only - Document what exists in code, no speculation
- Writing Tone - Clear and educational without audience labels
- Code Examples - Real files with paths and line numbers
- Performance Docs - Techniques + measurement methods, NOT estimated timings
- What NOT to include - No troubleshooting, future work, or meta-commentary
- Diagrams - Use when they clarify technicals, not for decoration
These principles override any template suggestions that conflict with them.
Note: Runbooks are the APPROPRIATE place for troubleshooting content (unlike feature docs).
Instructions
When documenting operational procedures:
- Identify the process (deploy, rollback, migration, etc.)
- Use the runbook template at
templates/runbook.md - Include actual commands that can be copy-pasted
- Document failure scenarios and recovery steps
- Output to
/docs/operations/[process-name].md
Template
Use the template at: .claude/skills/document-runbook/templates/runbook.md
Runbook Categories
Organize runbooks by category:
docs/operations/
├── index.md # Operations overview
├── deployment/
│ ├── deploy-to-production.md
│ ├── deploy-to-staging.md
│ └── rollback-deployment.md
├── database/
│ ├── database-migration.md
│ ├── backup-restore.md
│ └── seed-data.md
├── maintenance/
│ ├── dependency-updates.md
│ └── log-rotation.md
└── incident-response/
├── service-outage.md
└── data-corruption.md
Command Standards
- All commands must be copy-paste ready
- Use environment variables for secrets:
$DATABASE_URL - Include
--dry-runoptions where available - Show both successful and error outputs
Risk Levels
| Level | Definition | Review Required | | ------ | ------------------------------------ | --------------- | | Low | No data loss risk, easily reversible | None | | Medium | Potential service disruption | Team lead | | High | Data loss risk, hard to reverse | Team approval |
Output Location
| Category | Output Path |
| ----------- | ---------------------------------------------- |
| Deployment | /docs/operations/deployment/[name].md |
| Database | /docs/operations/database/[name].md |
| Maintenance | /docs/operations/maintenance/[name].md |
| Incident | /docs/operations/incident-response/[name].md |
Quality Checklist
Before completing:
- [ ] All commands are copy-paste ready
- [ ] Expected outputs documented
- [ ] Failure scenarios covered
- [ ] Rollback procedure included
- [ ] Troubleshooting table complete
- [ ] Prerequisites clearly listed
- [ ] Time estimate provided
- [ ] Risk level assessed
- [ ] Emergency contacts included (for high-risk)
Examples
Creating New Runbooks (Will Ask User)
- "Create a deployment runbook" → Ask: Risk level? Environment?
- "Document the database migration process" → Confirm category and scope
Updating Existing Runbooks (Autonomous)
- "Add new step to deployment runbook" → Updates existing doc
- "Fix the database restore command" → Corrects command
- "Add troubleshooting for timeout errors" → Adds to troubleshooting table
By Category
- "Document production deployment" →
/docs/operations/deployment/deploy-to-production.md - "Create database backup runbook" →
/docs/operations/database/backup-restore.md - "Write incident response for outages" →
/docs/operations/incident-response/service-outage.md
Architecte Docker Compose
DevOps
Concoit des configurations Docker Compose optimisees.
Rapport de Post-Mortem
DevOps
Rédige des rapports post-mortem d'incidents structurés et blameless.
Créateur de Runbooks
DevOps
Crée des runbooks opérationnels clairs pour les procédures DevOps courantes.