v1truv1us Default output: return only the result, blockers, and required evidence. Omit preambles, process narration, repeated context, confidence scores, and follow-up offers. Use at most five bullets unless a required artifact or schema needs more.
Playwright Test Engineering
Current Versions (Verify Before Use)
npx playwright --version
Check Playwright releases for the latest version.
Core Principles
- Tests should be deterministic. No random data, no time-based assertions, no shared state.
- Prefer user-visible selectors. Use
getByRole,getByText,getByLabelover CSS selectors. - One assertion per test. Test one thing. If it fails, you know exactly what broke.
- Traces are for debugging, not primary detection. A failing test should explain itself; traces help when it doesn't.
- Parallel by default. Use fully-parallel mode. Tests must be isolated.
Test Writing Checklist
- Uses semantic locators (
getByRole,getByLabel,getByText) - Avoids
waitForTimeout— uses explicit waits (waitForSelector,waitForResponse) - Each test has
test.describegrouping - Test data is isolated (no shared accounts, no hardcoded IDs)
- API calls are mocked or use test-specific data
- Screenshots/traces are captured on failure
Codegen Workflow
# Generate test from user interactions
npx playwright codegen http://localhost:3000
# Run with UI mode for debugging
npx playwright test --ui
# Run with trace viewer
npx playwright test --trace on
npx playwright show-trace test-results/trace.zip
CI Integration
# .github/workflows/playwright.yml
- uses: actions/checkout@v4
- uses: actions/setup-node@v4
with:
node-version: 22
- run: npm ci
- run: npx playwright install --with-deps
- run: npx playwright test
- uses: actions/upload-artifact@v4
if: failure()
with:
name: playwright-report
path: playwright-report/
Common Anti-Patterns
| Anti-Pattern | Why It's Wrong | Fix |
|---|---|---|
page.click('.btn-primary') |
Brittle, no semantic meaning | page.getByRole('button', { name: 'Submit' }).click() |
page.waitForTimeout(1000) |
Flaky, slow | page.waitForResponse('**/api/data') |
| Shared test account | State leaks between tests | Isolate data per test |
| No retries in CI | Network flakes cause failures | retries: 2 in CI |
| Screenshots on every run | Wastes storage | screenshot: 'only-on-failure' |
| Hardcoded URLs | Environment-specific breakage | Use baseURL config |
| No visual regression | UI changes break layout | Add expect(page).toHaveScreenshot() |
Validation Checklist
-
playwright.config.tshasfullyParallel: true -
retriesconfigured for CI (2 for CI, 0 for local) -
projectscover required browsers (Chromium minimum, +WebKit for Safari) - Trace configured:
trace: 'on-first-retry' -
globalSetuphandles test data seeding if needed -
testResultsDiris in.gitignore