E2e Playwright

Playwright E2E Testing

Production-tested patterns from the TestDino Playwright Skill. Every pattern includes when (and when not) to use it.

Golden Rules

1. getByRole() over CSS/XPath — resilient to markup changes, mirrors how users see the page
2. Never page.waitForTimeout() — use expect(locator).toBeVisible() or page.waitForURL()
3. Web-first assertions — expect(locator) auto-retries; expect(await locator.textContent()) does NOT
4. Isolate every test — no shared state, no execution-order dependencies
5. baseURL in config — zero hardcoded URLs in tests
6. Retries: 2 in CI, 0 locally — surface flakiness where it matters
7. Traces: 'on-first-retry' — rich debugging artifacts without CI slowdown
8. Fixtures over globals — share state via test.extend(), not module-level variables
9. One behavior per test — multiple related expect() calls are fine
10. Mock external services only — never mock your own app; mock third-party APIs, payment gateways, email
11. Real auth or stop — never test.skip(true, ...) around missing login, never .or(signIn) assertions that pass on the auth wall. If auth setup doesn't exist, STOP and set up storage state with the user (one-time). A test that passes signed-out is not a feature test.

Deep dives available in references/ directory — read them when working on the relevant topic.

Feature Tests vs Smoke Tests

Not all E2E tests are equal. Know what tier you're writing.

The rule: Every feature shipped MUST have at least one tier-2 (feature) E2E test. Smoke tests are free but DO NOT count toward feature coverage.

Ask yourself: "If someone broke this feature tomorrow, would my E2E tests catch it?" If the answer is "only if they deleted the entire page" — you wrote smoke tests, not feature tests.

Navigation Tests — Required When Nav Changes

When you add or modify navigation (sidebar items, mobile tab bar, header links, route changes), you MUST write tests that verify:

1. Nav item is visible at the correct viewport (desktop sidebar, mobile tab bar)
2. Clicking it navigates to the correct URL
3. Destination page renders its primary content (not just "no 404")
4. Active/selected state highlights correctly

Desktop + Mobile navigation test template:

import { test, expect } from '@playwright/test';

test.describe('Navigation — Desktop', () => {
  test.use({ viewport: { width: 1280, height: 800 } });

  test('sidebar contains Canvas link and navigates correctly', async ({ page }) => {
    await page.goto('/');
    const sidebar = page.getByRole('navigation');
    const canvasLink = sidebar.getByRole('link', { name: 'Canvas' });
    await expect(canvasLink).toBeVisible();
    await canvasLink.click();
    await page.waitForURL('/canvas');
    await expect(page.getByRole('heading', { name: 'Canvas' })).toBeVisible();
  });
});

test.describe('Navigation — Mobile', () => {
  test.use({ viewport: { width: 375, height: 812 } });

  test('mobile tab bar contains Canvas and navigates correctly', async ({ page }) => {
    await page.goto('/');
    const tabBar = page.getByRole('navigation', { name: /mobile|tab/i });
    const canvasTab = tabBar.getByRole('link', { name: 'Canvas' });
    await expect(canvasTab).toBeVisible();
    await canvasTab.click();
    await page.waitForURL('/canvas');
    await expect(page.getByRole('heading', { name: 'Canvas' })).toBeVisible();
  });
});

Adapt names/selectors to the actual app. The structure is: find nav → find link → click → verify URL → verify content.

Next.js Config (App Router + Pages Router)

// playwright.config.ts
import { defineConfig, devices } from '@playwright/test';

export default defineConfig({
  testDir: './tests/e2e',
  fullyParallel: true,
  forbidOnly: !!process.env.CI,
  retries: process.env.CI ? 2 : 0,
  workers: process.env.CI ? '50%' : undefined,
  reporter: process.env.CI ? 'html' : 'list',

  use: {
    baseURL: 'http://localhost:3000',
    trace: 'on-first-retry',
    screenshot: 'on',
    video: 'retain-on-failure',
  },

  expect: {
    toHaveScreenshot: {
      maxDiffPixelRatio: 0.01,
      animations: 'disabled',
    },
  },

  projects: [
    { name: 'chromium', use: { ...devices['Desktop Chrome'] } },
    { name: 'mobile', use: { ...devices['iPhone 14'] } },
  ],

  webServer: {
    command: process.env.CI
      ? 'npm run build && npm run start'  // production build in CI
      : 'npm run dev',                    // dev server locally
    url: 'http://localhost:3000',
    reuseExistingServer: !process.env.CI,
    timeout: 120_000,
    env: {
      NODE_ENV: process.env.CI ? 'production' : 'test',
    },
  },
});

Environment variables: Next.js loads .env.test automatically when NODE_ENV=test. Use .env.test for non-secret test config (committed), .env.test.local for secrets (gitignored).

Gitignore additions:

.env*.local
playwright-report/
playwright/.auth/
test-results/
blob-report/

Do NOT gitignore screenshot baselines. The *.spec.ts-snapshots/ directories created by toHaveScreenshot() MUST be committed — they are the source of truth for visual regression tests. Only ephemeral artifacts (test-results/, playwright-report/) should be ignored.

Locators — Priority Order

Use the first one that works:

page.getByRole('button', { name: 'Submit' })         // 1. Role (ALWAYS preferred)
page.getByLabel('Email address')                      // 2. Label (form fields)
page.getByText('Welcome back')                        // 3. Text (non-interactive content)
page.getByPlaceholder('Search...')                     // 4. Placeholder
page.getByAltText('Company logo')                     // 5. Alt text (images)
page.getByTitle('Close dialog')                       // 6. Title attribute
page.getByTestId('checkout-summary')                  // 7. Test ID (last resort)
page.locator('css=.legacy-widget')                    // 8. CSS/XPath (absolute last resort)

Role locator cheat sheet:

// Buttons — matches <button>, <input type="submit">, role="button"
page.getByRole('button', { name: 'Save changes' })

// Links — matches <a href>
page.getByRole('link', { name: 'View profile' })

// Headings — use level to target h1-h6
page.getByRole('heading', { name: 'Dashboard', level: 1 })

// Text inputs — by accessible name (label)
page.getByRole('textbox', { name: 'Email' })

// Checkboxes and radios
page.getByRole('checkbox', { name: 'Remember me' })
page.getByRole('radio', { name: 'Monthly billing' })

// Dropdowns — <select> elements
page.getByRole('combobox', { name: 'Country' })

// Navigation landmarks
page.getByRole('navigation', { name: 'Main' })

// Dialogs
page.getByRole('dialog', { name: 'Confirm deletion' })

// Exact matching — prevents "Log" matching "Log out"
page.getByRole('button', { name: 'Log', exact: true })

For deeper locator strategy guidance, read references/locators-deep-dive.md

Assertions — Web-First vs Non-Retrying

Web-first (auto-retry) — ALWAYS prefer:

await expect(page.getByRole('heading')).toBeVisible();
await expect(page.getByRole('heading')).toHaveText('Dashboard');
await expect(page.getByRole('listitem')).toHaveCount(5);
await expect(page.getByRole('button')).toBeEnabled();
await expect(page.getByLabel('Name')).toHaveValue('Jane');
await expect(page.getByTestId('card')).toHaveClass(/active/);
await expect(page.getByRole('checkbox')).toBeChecked();
await expect(page.getByRole('dialog')).not.toBeVisible();

Non-retrying — only for already-resolved values:

const title = await page.title();
expect(title).toBe('Health Check');

const response = await page.request.get('/api/users');
expect(response.status()).toBe(200);

Polling assertion — non-DOM async conditions:

await expect.poll(() => getUserCount()).toBe(10);

Retry block — multiple assertions that must pass together:

await expect(async () => {
  const count = await page.getByRole('listitem').count();
  expect(count).toBeGreaterThan(0);
}).toPass();

Critical mistake: expect(await locator.textContent()).toBe('x') — this resolves ONCE with no retry. Use await expect(locator).toHaveText('x') instead.

Visual Regression

When to use:

Quick reference:

// Full page baseline
await expect(page).toHaveScreenshot('homepage.png');

// Element-level (smaller, more stable)
await expect(page.getByTestId('nav')).toHaveScreenshot('nav.png');

// Full scrollable page
await expect(page).toHaveScreenshot('pricing.png', { fullPage: true });

// With masking for dynamic content
await expect(page).toHaveScreenshot('profile.png', {
  mask: [page.getByTestId('timestamp'), page.getByTestId('avatar')],
});

Baseline workflow:

# Generate baselines (first run or after intentional UI change)
npx playwright test --update-snapshots

# Commit baselines — they are the source of truth
git add tests/e2e/**/*.spec.ts-snapshots/
git commit -m "test: add/update Playwright screenshot baselines"

CRITICAL: Screenshot baselines MUST be committed. Without them, toHaveScreenshot() fails on the next run because there's nothing to compare against. Never gitignore *.spec.ts-snapshots/ directories.

For thresholds, CI consistency, masking strategies, and anti-patterns, read references/visual-regression-deep-dive.md

Authentication

If auth setup does not exist yet, STOP and create it with the user before writing feature tests. You cannot mint a test account yourself — ask for the test user credentials once, wire the storage-state pattern below, and every future test runs authenticated. Two patterns are banned outright:

// BANNED: the test permanently skips itself — it has never actually run
test.skip(!(await isSignedIn(page)), 'Not authenticated');

// BANNED: passes when the feature is completely inaccessible
await expect(
  page.getByRole('heading', { name: 'Sign in' })
    .or(page.getByRole('heading', { name: 'Canvas' }))
).toBeVisible();

Both report green while verifying nothing. (Audit result: one project shipped 25 "E2E tests" where 18 were skip-guarded and had never executed once.) Skipped is failing — write the auth setup instead.

Storage state reuse (default pattern):

// global-setup.ts — run once before all tests
import { chromium, type FullConfig } from '@playwright/test';

async function globalSetup(config: FullConfig) {
  const { baseURL } = config.projects[0].use;
  const browser = await chromium.launch();
  const context = await browser.newContext();
  const page = await context.newPage();

  await page.goto(`${baseURL}/login`);
  await page.getByLabel('Email').fill(process.env.TEST_USER_EMAIL!);
  await page.getByLabel('Password').fill(process.env.TEST_USER_PASSWORD!);
  await page.getByRole('button', { name: 'Sign in' }).click();
  await page.waitForURL('**/dashboard');
  await context.storageState({ path: '.auth/user.json' });
  await browser.close();
}
export default globalSetup;

// playwright.config.ts
export default defineConfig({
  globalSetup: require.resolve('./global-setup'),
  projects: [
    { name: 'setup', testMatch: /.*\.setup\.ts/ },
    {
      name: 'chromium',
      use: {
        ...devices['Desktop Chrome'],
        storageState: '.auth/user.json',
      },
      dependencies: ['setup'],
    },
  ],
});

Add .auth/ to .gitignore — auth state files contain session tokens.

For multi-role auth, API login, and NextAuth patterns, read references/authentication-deep-dive.md

Fixtures — Prefer Over Hooks

Rule: If it needs cleanup, use a fixture. If it doesn't and is simple, a hook is okay.

// fixtures.ts
import { test as base, expect } from '@playwright/test';

export const test = base.extend<{ todoPage: TodoPage }>({
  todoPage: async ({ page }, use) => {
    // Setup
    await page.goto('/todos');
    const todoPage = new TodoPage(page);
    await use(todoPage);    // Hand to test
    // Teardown — runs even if test crashes
    await page.evaluate(() => localStorage.clear());
  },
});

// Worker-scoped (expensive, shared across tests in one worker)
export const test = base.extend<{}, { dbConnection: DatabaseClient }>({
  dbConnection: [async ({}, use) => {
    const db = await DatabaseClient.connect(process.env.DB_URL!);
    await use(db);
    await db.disconnect();
  }, { scope: 'worker' }],
});

Network Mocking — External Services Only

Decision: Mock at the boundary, test your stack end-to-end.

// Mock a third-party payment API
await page.route('**/api/create-payment-intent', route =>
  route.fulfill({
    status: 200,
    contentType: 'application/json',
    body: JSON.stringify({ clientSecret: 'pi_mock_123', amount: 9900 }),
  })
);

// Block analytics entirely
await page.route('**/analytics.example.com/**', route => route.abort());

// Wait for a specific API response
const responsePromise = page.waitForResponse('**/api/users');
await page.getByRole('button', { name: 'Load' }).click();
await responsePromise;

For HAR recording, conditional mocking, and advanced patterns, read references/mocking-deep-dive.md

Flaky Test Diagnosis

Taxonomy — identify the category first:

Decision tree:

Fails locally with --repeat-each=20?
├── YES → TIMING issue: missing await, waitForTimeout, race condition
└── NO → Fails only in CI?
    ├── YES → ENVIRONMENT: viewport, fonts, slower machines, missing deps
    └── NO → Fails only with other tests?
        ├── YES → ISOLATION: shared state, DB leaks, localStorage
        └── NO → INFRASTRUCTURE: browser crash, OOM, DNS

Fixes for timing (most common):

// ❌ Arbitrary wait
await page.waitForTimeout(3000);
await expect(page.getByTestId('chart')).toBeVisible();

// ✅ Auto-retrying assertion
await expect(page.getByTestId('chart')).toBeVisible();

// ❌ Clicks without waiting for network
await page.getByRole('button', { name: 'Load More' }).click();
await expect(page.getByRole('listitem')).toHaveCount(20);

// ✅ Wait for API response first
const responsePromise = page.waitForResponse(
  resp => resp.url().includes('/api/users') && resp.status() === 200
);
await page.getByRole('button', { name: 'Load More' }).click();
await responsePromise;
await expect(page.getByRole('listitem')).toHaveCount(20);

// ❌ Click during animation
await page.getByRole('button', { name: 'Open' }).click();
await page.getByRole('button', { name: 'Confirm' }).click();

// ✅ Wait for dialog to stabilize
await page.getByRole('button', { name: 'Open' }).click();
await expect(page.getByRole('dialog')).toBeVisible();
await page.getByRole('button', { name: 'Confirm' }).click();

Stability validation:

# Burn-in: run 10 times to confirm stability
npx playwright test tests/checkout.spec.ts --repeat-each=10

# Run in isolation to rule out state leaks
npx playwright test -g "adds item" --workers=1

# Full parallel to expose isolation issues
npx playwright test --fully-parallel --workers=4

Debugging Workflow

Follow this order. Most issues resolve by step 2.

1. Read the full error message
   └─ Check references/common-pitfalls.md for known patterns
2. Run with --ui to see what happened visually
   └─ Timeline shows every action + screenshot at failure
3. Enable tracing: use: { trace: 'on' } temporarily
4. Check network tab in trace for API failures
   └─ Missing responses, 4xx/5xx, CORS
5. Insert page.pause() at failure point
   └─ Inspect live DOM, try selectors in console
6. Check browser console for JS errors
   └─ page.on('console') or console tab in trace

Commands:

npx playwright test --ui                           # Interactive UI mode
npx playwright test --headed                       # See browser
npx playwright test --headed --slow-mo=500         # Slow motion
PWDEBUG=1 npx playwright test tests/login.spec.ts  # Step-through inspector
npx playwright show-trace test-results/*/trace.zip # View CI trace
DEBUG=pw:api npx playwright test                   # Verbose API logs

ESLint rule to catch missing awaits:

{ "rules": { "@typescript-eslint/no-floating-promises": "error" } }

Common Pitfalls (Top 10)

For all 20 pitfalls with full code examples, read references/common-pitfalls.md

Self-audit before claiming E2E coverage

grep -rn   "test\.skip(true" tests/e2e && echo "FAIL: permanently skipped tests"
grep -rniE "sign.?in.*\.or\(|\.or\(.*sign.?in" tests/e2e && echo "FAIL: auth-wall tautology"
grep -rn   "waitForTimeout"  tests/e2e && echo "FAIL: arbitrary waits"

(The tautology grep matches both operand orders — signIn.or(content) and content.or(signIn) — because real offenders write it both ways.)

Any hit means the coverage claim is false. Fix the tests before proceeding — do not report them as passing.

Next.js Specific Patterns

App Router — server components render before Playwright sees the page:

test('home page renders server component', async ({ page }) => {
  await page.goto('/');
  // SSR content is already in HTML by the time Playwright loads
  await expect(page.getByRole('heading', { name: 'Welcome', level: 1 })).toBeVisible();
});

Loading states with streaming/suspense:

test('loading skeleton during data streaming', async ({ page }) => {
  // Slow the API to expose loading state
  await page.route('**/api/dashboard/stats', async route => {
    await new Promise(r => setTimeout(r, 2000));
    await route.continue();
  });
  await page.goto('/dashboard');
  await expect(page.getByRole('progressbar')).toBeVisible();
  await expect(page.getByRole('heading', { name: 'Stats' })).toBeVisible();
});

API routes:

test('API route returns expected data', async ({ request }) => {
  const response = await request.get('/api/users');
  expect(response.ok()).toBe(true);
  const data = await response.json();
  expect(data.users).toHaveLength(3);
});

Client-side navigation:

test('client-side navigation preserves state', async ({ page }) => {
  await page.goto('/dashboard');
  await page.getByRole('textbox', { name: 'Search' }).fill('test query');
  await page.getByRole('link', { name: 'Settings' }).click();
  await page.waitForURL('/settings');
  await page.getByRole('link', { name: 'Dashboard' }).click();
  await page.waitForURL('/dashboard');
  // Client navigation preserved — check if state survives
  await expect(page.getByRole('textbox', { name: 'Search' })).toHaveValue('test query');
});

For middleware testing, route groups, parallel routes, and NextAuth patterns, read references/nextjs-deep-dive.md

Reference Files

For deep dives, read the relevant file in references/: