TDD Playwright Workflow

Purpose

Make implementation evidence-driven: reproduce the behavior first, make the smallest fix, then prove the user workflow works through unit and e2e tests where appropriate.

Inspect First

• existing test runners, Playwright config, browser test directories, fixtures, package scripts, CI, and user workflow docs
• changed behavior, bug reports, and product acceptance criteria

Procedure

1. Choose the reproduction layer.

   • Unit test for pure logic, validation, formatting, reducers, and domain behavior.
   • Integration test for API, database, service collaboration, or contract behavior.
   • Playwright e2e for complete browser workflows such as create, edit, delete, checkout, login, or publish.

2. Write the failing test first.

   • For bug fixes, add the failing reproduction before changing production code unless impossible.
   • If impossible, document why in PR/final validation notes.
   • Use assets/templates/tdd-bug-reproduction.md.tmpl for notes.

3. Implement minimally.

   • Make the reproduction pass.
   • Avoid unrelated refactors during red/green.
   • Use typescript-monorepo-clean-code if cleanup is needed after behavior is green.

4. Add or verify Playwright.

   • Use existing Playwright setup when present.
   • Add Playwright only when the task accepts browser e2e tooling.
   • Use assets/templates/playwright-e2e.spec.ts.tmpl as a starting point.
   • Ensure scripts/test-e2e exists when e2e tests are part of the repo contract.
   • Keep unit and e2e discovery separate. In Bun repos, do not let bun test discover Playwright *.spec.ts files; narrow scripts/test to unit/integration paths or place Playwright files under an ignored e2e pattern.
   • Ignore Playwright output directories such as test-results and playwright-report in repo linters/formatters.
   • Use $click-path-and-browser-qa to inventory routes, controls, accessible names, and missing lifecycle coverage before broad UI test work.

5. Validate the loop.

   • Run the narrow failing test, then relevant broader tests.
   • Use chrome-devtools-runtime-debugger for runtime-only console, network, layout, or performance investigation.
   • Record completion evidence through termination-gatekeeper and feature evidence through feature-state-machine when present.

Validation

• Show red/green evidence when feasible.
• Run unit and/or Playwright commands that cover the change.
• Run ./scripts/test and ./scripts/validate when the change is broad.
• After running Playwright, rerun ./scripts/test or ./scripts/validate to confirm e2e output and file globs did not break baseline validation.

Completion Criteria

• Feature or bug behavior is covered at the narrowest useful layer.
• Browser lifecycle changes have Playwright or a documented reason e2e is not feasible.
• Validation evidence names exact commands and outcomes.