CI/CD Pipeline Builder Skill

Writing a CI/CD pipeline from scratch means reading documentation for an afternoon, iterating through YAML syntax errors, and eventually getting something that mostly works. This skill produces a working pipeline on the first pass — with caching, proper job dependencies, and the environment-specific logic that always gets left out of tutorials.

What the skill does

• Full pipeline generation. Test, lint, build, and deploy stages wired together with correct dependencies and failure handling.
• Dependency caching. Proper cache configuration for npm, pip, Maven, Gradle, or Go modules — the single biggest impact on pipeline speed.
• Matrix builds. Multi-version and multi-platform testing configurations.
• Environment-specific deployments. Staging deploys on push to main, production deploys on tags, with manual approval gates where needed.
• Secrets handling. Correct patterns for passing secrets to jobs without exposing them in logs.
• Optimisation. Parallel jobs, conditional steps, and cache strategies that cut pipeline time without cutting corners.
• Debugging. Diagnoses existing pipeline failures from error output and config.

How it works

1. Describe the project. Language, framework, how tests are run, how the app is deployed (Docker, serverless, static hosting, etc.), and any specific requirements.
2. Specify the platform. GitHub Actions, GitLab CI, CircleCI, Bitbucket Pipelines — the skill adapts the syntax.
3. Pipeline draft. A complete YAML file for the platform, ready to commit.
4. Iteration. Ask to add a stage, adjust the deployment logic, or optimise for speed.

How to use it

1. Click ⬇ Download this Claude Skill above.
2. Import through Claude Desktop (Customize → Skills → + → Create skill → Upload a skill) or drop into .claude/commands/. Full walkthrough in the import tutorial.
3. Invoke the skill:

   /cicd-builder
   

   Or specify the setup:

   /cicd-builder GitHub Actions. Node.js 20, tests with Jest, Docker build, deploy to Fly.io on main. Deploy to production only on version tags.
   

---

Quick-start prompt (no download)

Write a complete CI/CD pipeline configuration for [platform: GitHub Actions / GitLab CI / etc.].

The pipeline should:

1. Run on every push and pull request
2. Install dependencies with caching (so repeated runs skip the download)
3. Run linting and tests in parallel where possible
4. Build a production artifact (Docker image / compiled binary / static files)
5. Deploy to staging on merge to main
6. Deploy to production on version tags (vX.Y.Z), with a manual approval step

Include:

• Correct job dependency declarations (needs:)
• Proper secrets handling — never echo secrets to logs
• Failure notifications or status checks
• Comments explaining non-obvious configuration choices

Project: [language, framework, how tests are run] Deployment target: [where the app runs — Fly.io, AWS, GCP, VPS, etc.] Special requirements: [anything specific — monorepo, multiple services, specific Node/Python version matrix, etc.]

Tips:

• Ask for a minimal pipeline first if you are starting from scratch — get it green, then add complexity.
• Caching is the highest-leverage optimisation; ask the skill to double-check your cache key strategy if builds feel slow.
• If jobs are failing intermittently, paste the error output — flaky tests, race conditions, and network timeouts have distinct signatures in CI logs.
• For monorepos, ask for path-filtered triggers so each service's pipeline only runs when its own files change.

⚠ This skill has been tested and optimized for Claude. Results may vary with other AI assistants.