Writing Tests

Add meaningful test coverage with TDD enforcement.

Best for: Improving coverage, TDD practice
Time estimate: 10-30 minutes
Skills used: mk:qa-manual (E2E), mk:review (coverage audit)
Agents involved: tester, developer (if implementation needed)

Overview

MeowKit's TDD discipline is opt-in via --tdd / MEOWKIT_TDD=1. When enabled, tests must exist and fail BEFORE implementation. When disabled (default), tests are recommended but not gated — you may write them before, alongside, or after the code. This workflow covers both adding unit tests to existing code and generating E2E tests from user flows.

Unit tests (TDD mode — --tdd)

Step 1: Write failing tests first

In TDD mode, the tester writes tests that define expected behavior:

/mk:cook add input validation to user registration --tdd

The tester writes tests BEFORE the developer touches implementation:

// tests/user-registration.test.ts (tester creates this)
test('rejects email without @ symbol', () => {
  expect(validateEmail('invalid')).toBe(false);
});

test('rejects password shorter than 8 characters', () => {
  expect(validatePassword('short')).toBe(false);
});

These tests fail — the validation functions don't exist yet.

Step 2: Developer implements to pass

The developer writes the minimum code to make tests green.

Step 3: Refactor

Both tester and developer refine — tests still pass after each change.

E2E tests (Playwright)

Step 1: Generate from spec

/mk:qa-manual tasks/plans/260327-checkout.md --generate

The mk:qa-manual skill:

1. Navigates the app like a human tester
2. Records every interaction as Playwright TypeScript code
3. Detects test ID convention (data-testid or data-cy)
4. Generates feature folder structure:

tests/e2e/checkout-flow/
├── common/
│   ├── checkoutSelectors.ts      # All locators centralized
│   ├── checkoutAssertions.ts     # Reusable assertions
│   └── checkoutIntercepts.ts     # API mocks
└── checkout-flow.spec.ts         # One describe block

Step 2: Verify selectors exist

The skill greps the source to verify each getByTestId() actually exists:

grep -rn 'data-testid="checkout-btn"' src/ → Found ✓
grep -rn 'data-testid="place-order"' src/ → Found ✓

Missing selectors get a warning comment and fallback locator.

Step 3: Run the generated tests

npx playwright test tests/e2e/checkout-flow/

Coverage audit

The mk:review skill includes a test coverage audit (Step 4.75) that:

1. Traces code paths through your changes
2. Maps user flows to test scenarios
3. Identifies untested paths
4. Auto-generates tests for gaps

Next workflow

→ QA Testing — manual QA with structured reports