Testing Strategy

Testing Strategy

Goal: 100% Test Coverage at All Times

Every feature must ship with tests. No exceptions.

Test Pyramid

Unit Tests (Vitest)

• All utility functions in src/lib/
• All server actions — mock DB, test business logic
• All API routes — test request/response
• DB schema validation — ensure constraints work
• Coverage target: 100% lines, branches, functions

E2E Tests (Playwright)

• Every user flow — login, create project, submit proposal, vote, comment
• Responsive testing — mobile + desktop viewports
• Dark mode — visual consistency
• Error states — invalid input, network errors, auth failures
• Accessibility — axe-core integration

Adversarial Testing Model

┌──────────────┐     writes tests     ┌──────────────┐
│   Codex CLI  │ ──────────────────→  │  Test Suite  │
└──────────────┘                      └──────┬───────┘
                                             │
                                      must pass
                                             │
┌──────────────┐     writes code      ┌──────▼───────┐
│ Claude Code  │ ──────────────────→  │  Codebase    │
└──────────────┘                      └──────────────┘

• Codex writes tests first (TDD style)
• Claude Code writes implementation to pass tests
• Neither agent sees the other's "strategy"
• This catches assumptions and blind spots

Test Commands

# Unit tests
npm run test              # single run
npm run test:watch        # watch mode

# E2E tests (requires Docker running)
npm run test:e2e          # headless
npx playwright test --ui  # interactive

# Coverage report
npx vitest run --coverage

CI/CD Integration (Future)

• GitHub Actions on PR
• Must pass all tests before merge
• Coverage report posted as PR comment
• Playwright screenshots on failure

Test File Conventions

• Unit: tests/unit/<module>.test.ts
• E2E: tests/e2e/<feature>.spec.ts
• Fixtures: tests/fixtures/
• Test utils: tests/helpers/

Known Testing Challenges

1. SQLite in Docker — tests run against Docker container, not local
2. Email testing — use mock SMTP or test accounts
3. AI responses — mock LLM calls, test prompt construction
4. Real-time updates — test WebSocket/SSE connections