Overview
Generate production-ready CI/CD pipeline configurations for automated testing, building, and deploying applications on GitLab CI and CircleCI. This skill creates well-structured workflows with proper caching, matrix testing, environment separation, and deployment strategies. For GitHub Actions pipelines, use the github-actions skill.
Instructions
When a user asks to create or improve a CI/CD pipeline, follow these steps:
Step 1: Analyze the project
Detect the project type and requirements:
bash
# Determine language and framework
ls package.json pyproject.toml Gemfile go.mod Cargo.toml pom.xml build.gradle 2>/dev/null
# Check for existing CI config
ls .gitlab-ci.yml .circleci/config.yml 2>/dev/null
# Detect test commands
cat package.json | grep -A5 '"scripts"' 2>/dev/null
cat Makefile 2>/dev/null | grep -E "^test|^lint|^build"
Identify:
- Language/runtime: Node.js, Python, Go, Rust, Java
- Package manager: npm, pnpm, yarn, pip, poetry
- Test framework: Jest, Pytest, Go test, etc.
- Build output: Docker image, static site, binary, package
- Deploy target: AWS, Docker registry, npm registry, SSH server
Step 2: Choose the CI/CD platform
Default to GitLab CI if the repo is on GitLab. Use CircleCI if specified or if the project already has a .circleci/ directory.
Step 3: Generate the pipeline configuration
GitLab CI — Node.js example:
yaml
# .gitlab-ci.yml
stages:
- lint
- test
- build
- deploy
variables:
NODE_VERSION: "20"
.node-cache:
cache:
key: ${CI_COMMIT_REF_SLUG}
paths:
- node_modules/
lint:
stage: lint
extends: .node-cache
image: node:${NODE_VERSION}
script:
- npm ci
- npm run lint
test:
stage: test
extends: .node-cache
image: node:${NODE_VERSION}
script:
- npm ci
- npm test -- --coverage
coverage: '/All files.*\|.*\s+([\d\.]+)/'
artifacts:
reports:
coverage_report:
coverage_format: cobertura
path: coverage/cobertura-coverage.xml
build:
stage: build
extends: .node-cache
image: node:${NODE_VERSION}
script:
- npm ci
- npm run build
artifacts:
paths:
- dist/
only:
- main
GitLab CI — Docker build and deploy:
yaml
build-image:
stage: build
image: docker:24
services:
- docker:24-dind
variables:
DOCKER_TLS_CERTDIR: "/certs"
script:
- docker login -u $CI_REGISTRY_USER -p $CI_REGISTRY_PASSWORD $CI_REGISTRY
- docker build -t $CI_REGISTRY_IMAGE:$CI_COMMIT_SHA .
- docker push $CI_REGISTRY_IMAGE:$CI_COMMIT_SHA
only:
- main
deploy:
stage: deploy
image: alpine:latest
before_script:
- apk add --no-cache openssh-client
- eval $(ssh-agent -s)
- echo "$SSH_PRIVATE_KEY" | ssh-add -
script:
- ssh -o StrictHostKeyChecking=no $DEPLOY_USER@$DEPLOY_HOST "docker pull $CI_REGISTRY_IMAGE:$CI_COMMIT_SHA && docker-compose up -d"
when: manual
only:
- main
CircleCI — Node.js example:
yaml
# .circleci/config.yml
version: 2.1
orbs:
node: circleci/node@5
jobs:
lint-and-test:
docker:
- image: cimg/node:20.11
steps:
- checkout
- node/install-packages
- run: npm run lint
- run: npm test -- --coverage
- store_test_results:
path: test-results
- store_artifacts:
path: coverage
build:
docker:
- image: cimg/node:20.11
steps:
- checkout
- node/install-packages
- run: npm run build
- persist_to_workspace:
root: .
paths: [dist]
deploy:
docker:
- image: cimg/node:20.11
steps:
- attach_workspace:
at: .
- run: npx vercel deploy --prod --token $VERCEL_TOKEN
workflows:
build-and-deploy:
jobs:
- lint-and-test
- build:
requires: [lint-and-test]
- deploy:
requires: [build]
filters:
branches:
only: main
Examples
Example 1: GitLab CI for a Django API with Docker deployment
User request: "Create a GitLab CI pipeline for my Django app with Docker deployment"
Actions taken:
- Detected: Django 4.2, Poetry, Pytest, PostgreSQL dependency
- Created
.gitlab-ci.ymlwith lint, test (with Postgres service), build, deploy stages - Added Postgres service container for integration tests
- Configured Docker image build and push to GitLab Container Registry
Result:
Created: .gitlab-ci.yml
Stages: lint -> test -> build -> deploy
- lint: ruff + mypy type checking
- test: pytest with PostgreSQL 16 service container
- build: Docker image build, pushed to $CI_REGISTRY_IMAGE
- deploy: SSH deploy to production (manual trigger)
Required variables: DEPLOY_HOST, DEPLOY_USER, SSH_PRIVATE_KEY
Example 2: CircleCI for a Node.js monorepo
User request: "Set up CircleCI for my monorepo with separate test jobs per package"
Actions taken:
- Detected: pnpm workspace with 3 packages (api, web, shared)
- Created
.circleci/config.ymlwith parallel test jobs per package - Used path filtering to only run jobs for changed packages
- Added build and deploy workflow for the web package
Result:
Created: .circleci/config.yml
Jobs: test-api, test-web, test-shared, build-web, deploy-web
- Uses path filtering: only tests changed packages
- Shared dependency caching across jobs
- Deploy to Vercel on main branch only
Required env vars: VERCEL_TOKEN
Estimated run time: ~4 minutes (parallel jobs)
Guidelines
- Enable dependency caching to speed up runs. GitLab uses
cache:blocks; CircleCI uses orbs orsave_cache/restore_cache. - Use service containers for database tests (Postgres, Redis, etc.) rather than installing them in the job.
- Separate CI (runs on every push/MR) from CD (runs only on main/tags).
- Store secrets in CI/CD variables, never in pipeline files.
- Use
only/rulesin GitLab CI orfiltersin CircleCI to control when jobs run. - For monorepos, use path-based triggers to only run relevant pipelines.
- GitLab CI supports
extendsand YAML anchors for DRY configs — use them for shared job configurations. - CircleCI orbs encapsulate common patterns (Node, Python, Docker) — prefer orbs over manual setup.
- Add
when: manualin GitLab CI or approval jobs in CircleCI for production deploys to prevent accidental releases.