# MANDATORY Tool Restrictions

BANNED TOOLS — calling these is a skill violation:

• EnterPlanMode — BANNED. Do NOT call this tool. This skill has its own structured testing workflow. Execute it directly.
• ExitPlanMode — BANNED. You are never in plan mode.</tool_restrictions>

Testing Strategy Workflow

Create comprehensive test strategies covering the full test pyramid. Execute with specialist agents.

**Read before planning:** 1. `${CLAUDE_PLUGIN_ROOT}/references/testing-patterns.md` — Test philosophy, vitest/playwright patterns 2. `${CLAUDE_PLUGIN_ROOT}/rules/testing.md` — Project conventions 3. `${CLAUDE_PLUGIN_ROOT}/references/llm-api-testing.md` — If testing LLM integrations 4. `${CLAUDE_PLUGIN_ROOT}/disciplines/change-impact-testing.md` — Blast radius analysis for code changes </required_reading>

Agents

This skill uses 3 writers + 2 runners:

Agent	Model	Purpose	Framework
unit-test-writer	sonnet	Write unit tests	vitest
integration-test-writer	sonnet	Write integration tests (API, auth)	vitest + MSW
e2e-test-writer	opus	Write E2E tests	Playwright
test-runner	haiku	Run vitest, analyze failures	vitest
e2e-runner	opus	Run Playwright, fix issues, iterate	Playwright

Test Pyramid:

         /\
        /  \       E2E (few)
       /────\      - Critical user journeys
      /      \     - Auth flows
     /────────\    Integration (some)
    /          \   - API interactions
   /────────────\  - Component + state
  /              \ Unit (many)
 /────────────────\- Pure functions
                   - Isolated components

**Check for project testing rules:**

Use Glob tool: .ruler/testing.md

If exists, read for MUST/SHOULD/NEVER constraints.

Detect test framework:

File	Framework
vitest.config.*	vitest (unit + integration)
playwright.config.*	Playwright (E2E)
</rules_context>

Process

Step 1: Determine Intent

"What would you like to do?"

Intent	Action
Create test strategy	Full test plan → dispatch agents
Run tests	Execute with appropriate runner
Fix failing tests	Dispatch debugger or e2e-runner
Review coverage	Analyze gaps and recommend

Step 2: Understand What's Being Tested

Gather context:

1. What feature or component?
2. Does it have auth? (Clerk, WorkOS, custom)
3. Does it call APIs?
4. Does it have critical user flows?
5. What's the risk level?

Step 3: Create Test Plan

For each feature, plan across all levels:

## Test Plan: [Feature Name]

### Risk Assessment
- **Criticality**: [high/medium/low]
- **Has Auth**: [yes/no — Clerk/WorkOS/custom]
- **Has API Calls**: [yes/no]
- **User-Facing**: [yes/no]

### Unit Tests (vitest)
**Agent:** unit-test-writer

| Test Case | What it Verifies |
|-----------|------------------|
| `should [behavior]` | [Pure function/component behavior] |
| `should [behavior]` | [Edge case handling] |
| `should [behavior]` | [Error throwing] |

**Files to create:**
- `src/[path]/[module].test.ts`

### Integration Tests (vitest + MSW)
**Agent:** integration-test-writer

| Test Case | What it Verifies |
|-----------|------------------|
| `should [behavior]` | [Component + API interaction] |
| `should [behavior]` | [Auth state handling] |
| `should [behavior]` | [Error from API] |

**Mocking required:**
- API endpoints: [list]
- Auth: [Clerk/WorkOS mock setup]

**Files to create:**
- `src/[path]/[feature].integration.test.ts`

### E2E Tests (Playwright)
**Agent:** e2e-test-writer

| Test Case | What it Verifies |
|-----------|------------------|
| `should [complete flow]` | [Happy path journey] |
| `should [handle error]` | [User-visible error] |
| `should [auth flow]` | [Login/logout if applicable] |

**Auth setup required:**
- Provider: [Clerk/WorkOS/none]
- Test user: [env var names]

**Files to create:**
- `tests/[feature].spec.ts`
- `tests/auth.setup.ts` (if auth needed)

Step 4: Execute Test Plan

Dispatch specialists in order:

1. Unit tests first (fastest feedback):

Task [unit-test-writer] model: sonnet: "Write unit tests for [feature].

Test cases from plan:
[paste unit test cases]

Files to create: [paths]
Follow vitest patterns from testing-patterns.md"

2. Integration tests second:

Task [integration-test-writer] model: sonnet: "Write integration tests for [feature].

Test cases from plan:
[paste integration test cases]

Auth: [Clerk/WorkOS/none]
API mocking: [endpoints to mock]
Files to create: [paths]"

3. E2E tests last:

Task [e2e-test-writer] model: opus: "Write E2E tests for [feature].

Test cases from plan:
[paste e2e test cases]

Auth: [Clerk/WorkOS/none]
Fixtures needed: [list]
Files to create: [paths]"

Step 5: Run and Verify

Unit + Integration (inline):

pnpm vitest run

E2E (background agent — avoids terminal issues):

Task [e2e-runner] model: opus: "Run E2E tests and fix any failures.

Test files: [list]
Iterate until green or report blockers."

---

Auth Testing Quick Reference

Clerk Testing

Integration tests (vitest):

• Mock useAuth and useUser hooks
• Test loading, signed in, signed out states
• Mock getToken for API calls

E2E tests (Playwright):

• Create tests/auth.setup.ts for login flow
• Store session in playwright/.auth/user.json
• Use storageState in playwright.config.ts

Common issues:

• ❌ Trying to mock ClerkProvider (mock hooks instead)
• ❌ Not handling isLoaded: false state
• ❌ Hardcoding tokens (use getToken mock)

WorkOS Testing

Integration tests (vitest):

• Mock getUser from @workos-inc/authkit-nextjs
• Test with full user object including organizationId, role, permissions
• Test SSO redirect behavior

E2E tests (Playwright):

• SSO flows are slow — consider test bypass endpoint
• Create /api/auth/test-login for faster auth (test env only)
• Store session state after auth

Common issues:

• ❌ Forgetting organizationId in mock (required for org-level features)
• ❌ Not testing permission checks
• ❌ SSO redirect timing issues (add proper waits)

Bypass Auth for Speed

For faster E2E tests, create a test-only auth endpoint:

// app/api/auth/test-login/route.ts
// ONLY available in test/development
export async function POST(request: Request) {
  if (process.env.NODE_ENV === "production") {
    return new Response("Not found", { status: 404 });
  }
  // Create session directly without SSO flow
}

---

Test Patterns

What to Test at Each Level

Level	Test	Don't Test
Unit	Pure functions, isolated components, hooks	API calls, multi-component flows
Integration	Component + API, auth states, form submissions	Full user journeys
E2E	Critical paths, auth flows, checkout/signup	Every possible input

Coverage Guidelines

Feature Type	Unit	Integration	E2E
Utility functions	✅ heavy	❌ none	❌ none
UI components	✅ rendering	✅ with state	❌ only if critical
Forms	✅ validation	✅ submission	✅ critical forms
API routes	✅ handlers	✅ with mocking	❌ via E2E
Auth flows	❌ none	✅ mock states	✅ real flow
Checkout/payment	✅ calculations	✅ flow	✅ full journey

Fail-Fast Configuration

Tests must fail fast. Never:

• Global timeout of minutes
• 5+ retries to mask flakiness
• Arbitrary sleeps

Playwright config:

export default defineConfig({
  timeout: 30_000,        // 30s max per test
  expect: {
    timeout: 5_000,       // 5s for assertions
  },
  use: {
    actionTimeout: 10_000, // 10s per action
  },
});

---

After creating test strategy or running tests:

## YYYY-MM-DD HH:MM — /arc:testing
**Task:** [Create strategy / Run tests / Fix failing]
**Feature:** [What was tested]
**Coverage:**
- Unit: [N] tests
- Integration: [N] tests
- E2E: [N] tests
**Auth:** [Clerk/WorkOS/none]
**Result:** [All passing / X failing]
**Next:** [Fix failures / Done]

---

</progress_append> Test strategy is complete when: - [ ] Risk assessment done - [ ] Unit test cases identified - [ ] Integration test cases identified (with mocking plan) - [ ] E2E test cases identified (with auth setup if needed) - [ ] Specialist agents dispatched - [ ] All tests written and passing - [ ] Coverage gaps noted </success_criteria>