CI/CD with GitHub Actions

I used to think CI/CD pipelines needed to be complex. Multiple stages, matrix builds, artifact caching, deployment gates. In practice, most projects just need: install dependencies, run tests, build. Everything else is optional until it isn’t.

Here’s the workflow I copy into new projects:

name: CI

on:
 push:
 branches: [main]
 pull_request:
 branches: [main]

jobs:
 build:
 runs-on: ubuntu-latest
 steps:
 - uses: actions/checkout@v4
 - uses: actions/setup-node@v4
 with:
 node-version: "20"
 cache: "npm"
 - run: npm ci
 - run: npm test
 - run: npm run build

That’s it. npm ci instead of npm install because it’s faster and respects the lockfile. Node 20 because 14 and 16 are dead. actions/cache is built into setup-node now, so you don’t need a separate cache step.

What I Add Later

Lint check - npm run lint or npx eslint . before tests. Catches style issues before review.
Type check - npx tsc --noEmit for TypeScript projects. Faster than a full build and catches type errors tests miss.
Deploy on tags - a separate job that only runs on v* tags, building and pushing to the hosting platform.

What I Don’t Add

Matrix builds across 4 Node versions for a side project. You’re not publishing a framework.
Complex artifact pipelines when npm run build takes 30 seconds.
Deployment from main branch pushes. I deploy from tags so I can control when releases go out.

The Real Value

The biggest benefit of CI isn’t the automation - it’s the social contract. When tests pass on CI, it means the code works in a clean environment, not just on your machine. When they fail, everyone sees it. That accountability matters more than any tooling choice.