e2e-testing

E2E Testing

Delegation to Core Test-Runner

This skill delegates execution to the playwright-coding-agent core test-runner skill with E2E-specific flags applied. Do not run npx playwright test directly from a spec — always route through the core runner so guardrails (budget monitoring, audit logging) are honoured.

Default Flags

Apply these flags for all E2E runs unless the caller explicitly overrides them:

Flag	Value	Purpose
--project	chromium	Run against Chromium by default; extend to firefox and webkit for cross-browser CI
--trace	on-first-retry	Capture a full trace on first retry to aid debugging without slowing green runs
--video	on-first-retry	Record video on first retry for visual regression investigation
--reporter	html	Generate the interactive HTML report; also emit list for CI log readability

Example invocation produced by this skill:

npx playwright test --project=chromium --trace on-first-retry --video on-first-retry --reporter=html,list

Running a Subset

When the caller specifies a file, directory, or tag, scope the run accordingly:

# Single spec file
npx playwright test tests/e2e/specs/login.spec.ts --project=chromium --trace on-first-retry

# Directory
npx playwright test tests/e2e/specs/checkout/ --project=chromium --trace on-first-retry

# Tagged tests (using Playwright tag annotations)
npx playwright test --grep @smoke --project=chromium --trace on-first-retry

HTML Report Parsing

After a run, open the HTML report with:

npx playwright show-report

Parse the generated playwright-report/index.html to extract:

1. Pass / fail / flaky counts — report as a structured summary.
2. Failed test names and file paths — list each failed test with its spec file location.
3. Retry information — note how many tests passed on retry (flaky candidates).

Trace Viewer URL Extraction

Traces are saved under test-results/<test-name>/trace.zip. For each failed test, extract and surface the trace viewer command:

npx playwright show-trace test-results/<test-name>/trace.zip

When reporting results, include the exact show-trace command for every failed test so the developer can open it immediately.

Failure Categorization

After parsing the report, classify failures into these categories and present them grouped:

Category	Indicators	Suggested action
Locator not found	locator.click: Target closed / waiting for locator timeout	Review selector with e2e-selectors skill
Assertion failed	expect(...).toBeVisible / toHaveText failure	Check expected vs actual in trace; review assertion
Navigation error	page.goto timeout / net::ERR_CONNECTION_REFUSED	Verify baseURL in .playwright-agent.json and app is running
Auth failure	Redirect to /login mid-test	Regenerate storageState via auth.setup.ts
Flaky	Passes on retry	Add retry logic in config or investigate race condition
Setup failure	Error in auth.setup.ts or global setup	Run setup project in isolation to isolate the issue

Reporting Format

Return a structured summary after every run:

E2E Run Summary
───────────────
Project:  chromium
Passed:   42
Failed:   2
Flaky:    1
Skipped:  0

Failed Tests:
  1. Login > should redirect to dashboard (tests/e2e/specs/login.spec.ts:14)
     Category: Assertion failed — expected URL /dashboard, received /login
     Trace:    npx playwright show-trace test-results/login-should-redirect/trace.zip

  2. Checkout > should apply promo code (tests/e2e/specs/checkout.spec.ts:88)
     Category: Locator not found — waiting for getByLabel('Promo code')
     Trace:    npx playwright show-trace test-results/checkout-promo/trace.zip

Flaky Tests:
  1. Dashboard > should load widgets (tests/e2e/specs/dashboard.spec.ts:31) — passed on retry