How to use the DevOps Guidelines Checklist agent skill

What the skill makes the agent do

When you ask for help with how software gets deployed, the agent follows the guide's advice. The skill covers deployment strategy (rolling, blue-green and canary), Docker, Kubernetes probes, CI/CD stages, health and shutdown behaviour, configuration through environment variables, rollback, feature flags, pre- and post-deploy checklists, and mobile release workflows with EAS and Fastlane.

The guide says to apply only the sections relevant to your workload, provider, environment and acceptance criteria. The agent skips the Docker, Kubernetes, mobile, rollback and other checks that do not fit your situation.

• Pinned base-image tags instead of :latest, with multi-stage builds and a non-root user
• Health endpoints such as GET /health, and graceful handling of SIGTERM where needed
• Canary or blue-green rollouts instead of replacing everything at once
• Environment variables validated at startup, with no committed secrets
• Rollback steps for Kubernetes, Vercel, Docker and mobile releases

How to install the gem-devops-guidelines skill

Download the skill ZIP from this page. The skill ships as a single file, SKILL.md, inside the archive. Unzip it into one of two locations, depending on how you want to use it.

For a personal setup available across your projects, unzip it into ~/.claude/skills/. For a setup shared with one repository, unzip it into the .claude/skills/ folder inside that project.

If you use the Claude apps, you can instead upload the ZIP under Settings → Capabilities → Skills.

• Personal: ~/.claude/skills/
• Project only: .claude/skills/ inside your project folder
• Claude apps: Settings → Capabilities → Skills, then upload the ZIP

How to trigger the skill with natural requests

You do not need a special command. Describe the task in plain language that mentions deployment, containers, pipelines, releases or production readiness. The skill's description covers DevOps, platform, container, pipeline and release tasks, so the agent is most likely to use it when your request matches those topics.

Be specific about your situation. The guide asks the agent to tailor advice to your workload, provider, environment and acceptance criteria, so including those details gives better answers.

• Review this Dockerfile before we ship it to production.
• Design a canary rollout for our API on Kubernetes.
• Write a production-readiness checklist for our service.
• Check our CI/CD pipeline stages and suggest what is missing.
• Draft a rollback plan for this Vercel deployment.
• Help me prepare an iOS release with EAS and TestFlight.

What the skill needs and how it behaves

The skill lists no requirements. It does not run scripts and does not use the network. It is guidance only: it contains no tools that act on your systems, so the agent does not deploy, change infrastructure or run commands for you.

Because it only supplies guidance, the output is advice and review notes. You or your team still carry out any change, and you should check the agent's suggestions against your own tooling and environment.

• Requirements: none listed
• Runs scripts: no
• Uses network: no
• Changes live systems: no, it is guidance only

Safety notes for using the skill

The skill is a reference, not an audit. Its own checklist is generic. For complex or regulated environments, or any situation that needs a full security audit, treat the output as a starting point and involve expert review.

Never paste live secrets into a prompt. The guide itself says secrets must never be committed and belong in environment variables or secret storage, so keep credentials out of the conversation as well as out of Git.

• Do not expect the agent to deploy or modify live infrastructure.
• Review every suggested command before running it.
• Keep credentials out of prompts and out of Git.

Rollback commands the guide references

When you ask about rollback, the guide points to these commands. Check them against your own setup before running anything, because the agent only describes them.

For Kubernetes it names kubectl rollout undo, for Vercel vercel rollback, and for Docker it says to redeploy the previous pinned image. For mobile, EAS users can roll back updates with eas update:rollback.