E2E Test Loop - Browser Automation Testing

Current branch: !git branch --show-current 2>/dev/null || echo "not in git repo"

The E2E test loop uses a 2-phase workflow with Dex task tracking for persistent, cross-session browser test development.

The 2-Phase Approach

Phase	Name	Purpose
1	Flow Analysis	Identify critical user flows
2	Dex Handoff	Create epic + tasks per flow

After Phase 2, use /complete <task-id> for each E2E test task.

Starting the Loop

/e2e "Cover checkout flow"            # Basic
/e2e "Test auth and settings flows"   # Multiple flows

Phase 1: Flow Analysis

1. Analyze application routes, features, user journeys
2. Identify critical flows needing E2E coverage
3. Prioritize 3-7 test tasks

Focus on:

• Happy paths users depend on
• Payment/auth/data submission flows
• Flows that broke in production

Output: <phase_complete phase="1"/>

Phase 2: Dex Handoff

Create Dex epic, then tasks for each flow:

# Create epic
dex create "E2E Test Coverage" --description "Critical user flow coverage"

# For each flow
dex create "E2E: checkout flow" --parent <epic-id> --description "
Flow: Browse → Cart → Checkout → Confirmation

Steps:
1. Add product to cart
2. Proceed to checkout
3. Fill payment form
4. Complete purchase

Files:
- e2e/checkout.e2e.page.ts
- e2e/checkout.e2e.ts

Acceptance:
- [ ] Page object with semantic locators
- [ ] Test covers happy path
"

Output: <phase_complete phase="2"/> or <promise>E2E SETUP COMPLETE</promise>

Working on Tasks

Use Dex + /complete workflow:

dex list --pending      # See what's ready
dex start <id>          # Start working
/complete <id>          # Run reviewers and complete

File Naming Convention

e2e/
├── checkout.e2e.page.ts    # Page object (locators, setup, actions)
├── checkout.e2e.ts         # Test file (concise tests)
├── auth.e2e.page.ts
└── auth.e2e.ts

Playwright Patterns

Locator Priority (Semantic First)

Priority	Locator	Example
1	getByRole	page.getByRole('button', { name: 'Submit' })
2	getByLabel	page.getByLabel('Email address')
3	getByText	page.getByText('Welcome back')
4	getByTestId	page.getByTestId('submit-btn') - last resort

Page Object Pattern

// checkout.e2e.page.ts
export class CheckoutPage {
  constructor(private page: Page) {}

  // Locators
  readonly emailInput = this.page.getByLabel('Email')
  readonly submitButton = this.page.getByRole('button', { name: 'Complete' })

  // Actions
  async fillEmail(email: string) {
    await this.emailInput.fill(email)
  }

  async submit() {
    await this.submitButton.click()
  }
}

// checkout.e2e.ts
test('user can complete checkout', async ({ page }) => {
  const checkout = new CheckoutPage(page)
  await checkout.fillEmail('user@example.com')
  await checkout.submit()
  await expect(page.getByText('Order confirmed')).toBeVisible()
})

Waiting Patterns

// Auto-waiting (built into actions)
await page.getByRole('button').click()  // Waits until clickable

// Explicit wait for state
await expect(page.getByText('Loaded')).toBeVisible()

// Wait for network
await page.waitForResponse('**/api/data')

Quality Standards

• ONE test per task - Focused, reviewable commits
• Test user-visible behavior - Not implementation details
• Tests must be independent - No shared state between tests
• Use page objects - Keep test files concise

Command Reference

/e2e "prompt"             # Start flow analysis
/cancel-e2e               # Cancel loop
/complete <task-id>       # Complete task with reviewers

Related

• /complete - Run reviewers and mark Dex task complete
• dex list - View pending tasks
• dex-workflow skill - Full Dex usage patterns