GitHub Actions is my go-to for QA pipelines. It’s free for public repos, tightly integrated with PRs, and YAML-based workflows are easy to version and review.
Basic test workflow
name: Test Suite
on:
pull_request:
branches: [main]
push:
branches: [main]
jobs:
test:
runs-on: ubuntu-latest
steps:
- uses: actions/checkout@v4
- name: Set up Python
uses: actions/setup-python@v5
with:
python-version: '3.12'
- name: Install dependencies
run: pip install -r requirements.txt
- name: Run tests
run: pytest --tb=short --junitxml=results.xml
- name: Upload results
uses: actions/upload-artifact@v4
if: always()
with:
name: test-results
path: results.xml
The if: always() is critical — you want artifacts even when tests fail.
Parallelizing with matrix
strategy:
matrix:
browser: [chromium, firefox, webkit]
steps:
- name: Run E2E tests
run: npx playwright test --project=${{ matrix.browser }}
Three browsers run in parallel. Total CI time stays the same as running one.
PR status checks
In your repo settings → Branch protection rules, require the test job to pass before merging. This makes broken tests a merge blocker — not a “we’ll fix it later.”
Caching dependencies
- uses: actions/cache@v4
with:
path: ~/.cache/pip
key: ${{ runner.os }}-pip-${{ hashFiles('requirements.txt') }}
Saves 30-60 seconds per run. Small thing, adds up across a team.
Notifications
Add a Slack notification step at the end to alert the team on main branch failures — not on every PR, that gets noisy fast.