ci-cd

Create a workflow triggered on push and pull_request, use actions/setup-python with a version matrix, enable pip caching, install dependencies, and run pytest. Keep the pipeline under 10 minutes by caching dependencies and parallelizing jobs.

Use actions/cache@v4 with a path like ~/.cache/pip and a key hashing your requirements or lock files, such as ${{ hashFiles('**/requirements.txt') }}. The setup-python action also supports built-in caching via the cache: 'pip' option.

No, CI (testing) and CD (deployment) should be separate workflows. CI runs on every push and pull request, while CD triggers only on merges to main, often with environment protection rules and manual approval for production.

Store secrets in GitHub repository or environment secrets and reference them with ${{ secrets.NAME }} in workflow env blocks. Never hardcode credentials in workflow files, and avoid echoing secret values into logs.

Use reusable workflows with the workflow_call trigger when multiple workflows duplicate the same jobs, such as a shared test job across repositories or Python versions. Call them with the uses keyword and pass inputs like python-version.