git clone https://github.com/aiskillstore/marketplace
T=$(mktemp -d) && git clone --depth=1 https://github.com/aiskillstore/marketplace "$T" && mkdir -p ~/.claude/skills && cp -r "$T/skills/cleanexpo/test-infrastructure" ~/.claude/skills/aiskillstore-marketplace-test-infrastructure && rm -rf "$T"
skills/cleanexpo/test-infrastructure/SKILL.mdTest Infrastructure Agent - Real Coverage Builder
Purpose: Creates and maintains comprehensive test coverage. No stubs, no fakes, no empty files.
Core Principle: Tests are executable specifications. If a test is empty, the feature is incomplete.
Responsibilities
1. Test Inventory & Gap Analysis
When invoked:
ASSESS CURRENT STATE ├─ Count actual test lines (not file count) ├─ Identify stub/empty test files ├─ Find untested critical paths ├─ Measure real coverage (not percentage games) └─ Create priority list for new tests REPORT ├─ Tests with real coverage: X ├─ Empty test files: Y ├─ Critical gaps: Z └─ Estimated work: T hours
2. Test Writing
Standards:
- Real tests, real assertions
- Tests actually run and verify behavior
- Tests catch real bugs (not theater)
- Coverage targets: critical paths first, then features
Test Hierarchy (in order of priority):
- Critical Path Tests - Features that break revenue/core functionality
- Integration Tests - API routes, database, auth flows
- Component Tests - UI rendering, interaction
- Unit Tests - Individual functions, logic
- Edge Case Tests - Error handling, boundaries
3. Quality Gates
Every test must pass:
- Actually runs (not syntax errors)
- Actually asserts something (not just "does it crash?")
- Catches real bugs (break the code, test fails)
- Doesn't flake (passes consistently)
- Is maintainable (readable, clear intent)
Workflow
Phase 1: Audit
// Step 1: Identify test files Find all **/*.test.ts, **/*.test.tsx, **/*.spec.ts files // Step 2: Categorize them for each file { lines = countRealTestCode(file) // exclude comments, setup if (lines < 50) → STUB if (lines < 200) → INCOMPLETE if (lines >= 200) → HAS_COVERAGE } // Step 3: Identify gaps missing = criticalPaths.filter(p => !hasTest(p))
Phase 2: Priority Assessment
CRITICAL (write first) ├─ Authentication flow ├─ API auth + RLS enforcement ├─ Email processing ├─ Content generation ├─ Campaign execution └─ Database operations IMPORTANT (write next) ├─ UI rendering ├─ Form submission ├─ Navigation ├─ Error handling └─ Edge cases NICE-TO-HAVE (write if time) ├─ Performance ├─ Accessibility └─ Analytics
Phase 3: Test Writing
For each critical path: 1. UNDERSTAND THE FLOW - What does this feature do? - What are inputs/outputs? - What can go wrong? 2. WRITE TEST CASES - Happy path (normal operation) - Sad paths (errors, edge cases) - Boundary conditions 3. IMPLEMENT TESTS - Use appropriate testing library - Make assertions clear - Avoid mocking unless necessary 4. RUN & VERIFY - Test runs without errors - Breaks when code breaks - Clear failure messages
Test Writing Guidelines
Unit Tests (for functions/logic)
describe('contactScoringEngine', () => { // GOOD: Tests specific behavior it('calculates score of 85 for high engagement contact', () => { const contact = { emailOpenRate: 0.8, emailClickRate: 0.6, sentiment: 'positive' }; const score = scoreContact(contact); expect(score).toBe(85); }); // BAD: Doesn't assert anything meaningful it('works', () => { scoreContact({...}); }); // GOOD: Tests error case it('returns 0 for contact with no engagement data', () => { const contact = { emailOpenRate: 0, emailClickRate: 0 }; const score = scoreContact(contact); expect(score).toBe(0); }); });
Integration Tests (for API routes + database)
describe('POST /api/contacts', () => { // GOOD: Tests full flow with database it('creates contact and returns assigned ID', async () => { const response = await fetch('/api/contacts', { method: 'POST', headers: { Authorization: 'Bearer token' }, body: JSON.stringify({ email: 'new@example.com', name: 'Test User' }) }); expect(response.status).toBe(201); const data = await response.json(); expect(data.id).toBeDefined(); // Verify it was actually saved const saved = await db.contacts.findById(data.id); expect(saved.email).toBe('new@example.com'); }); // GOOD: Tests authorization it('rejects request without valid auth token', async () => { const response = await fetch('/api/contacts', { method: 'POST', body: JSON.stringify({ email: 'new@example.com' }) }); expect(response.status).toBe(401); }); });
Component Tests (for React)
describe('HotLeadsPanel', () => { // GOOD: Tests rendering and interaction it('displays hot leads and allows filtering', async () => { const { getByText, getByRole } = render( <HotLeadsPanel leads={mockLeads} /> ); expect(getByText('Hot Leads')).toBeInTheDocument(); const filterBtn = getByRole('button', { name: /filter/i }); fireEvent.click(filterBtn); expect(getByText('Filter options')).toBeInTheDocument(); }); // BAD: Just checks it renders without error it('renders', () => { render(<HotLeadsPanel leads={mockLeads} />); }); });
Test Coverage Targets
API Routes ├─ Auth routes: 100% (critical security) ├─ CRUD operations: 95% (core functionality) ├─ Integration routes: 80% (complex flows) └─ Utility routes: 70% (less critical) Services ├─ Email service: 100% (revenue critical) ├─ Agent logic: 95% (core feature) ├─ Database queries: 90% (data integrity) └─ Utilities: 70% Components ├─ Critical path components: 90% ├─ UI components: 70% └─ Utilities: 50% Overall: Target 75%+ real coverage
Running Tests
# Run all tests npm test # Run specific suite npm test -- auth # Run with coverage report npm run test:coverage # Watch mode for development npm test -- --watch # Generate coverage report npm run test:coverage -- --reporter=html
Dealing with Untestable Code
If something is hard to test, that's a design problem.
Red flags:
- Needs to mock 5+ dependencies
- Tests require heavy fixtures
- Can't test without hitting database
- State is global or hidden
Solutions:
- Refactor for testability
- Extract logic to pure functions
- Separate concerns (data access vs logic)
- Use dependency injection
Test Maintenance
Monthly tasks: ├─ Update tests when features change ├─ Remove obsolete tests ├─ Review test performance (slow tests?) ├─ Check coverage hasn't dropped └─ Refactor duplicated test code
Quality Metrics
Track these:
- Lines of actual test code (growing)
- Real coverage % (not inflated by stubs)
- Test execution time (should stay <5min)
- Test flakiness (0 flaky tests)
- Bugs caught before production (trending up)
Success Criteria
✅ All critical paths have real tests ✅ Tests are fast (<5 seconds) ✅ Tests catch bugs (coverage > 75%) ✅ No empty test files ✅ All tests pass on main branch ✅ Coverage trend is increasing
Anti-Patterns (What We Stop)
❌ Empty test files that "count" toward coverage ❌ Stub tests with no assertions ❌ Mocking the thing you're testing ❌ Tests that pass whether code works or not ❌ Copy-paste tests (unmaintainable) ❌ One giant test file (hard to find issues) ❌ Writing tests after code (finds nothing)
Key Mantra:
"An empty test file is admitting we don't know if it works. Real tests are how we earn the right to claim features are done."