Playwright

@varianz/playwright drives a real browser against an instrumented backend: the fixture creates a session per test and automatically sets the x-varianz-id header on the page, so every request the browser makes carries your session — while you override and probe VPoints anywhere in the backend graph. Install it (instructions) and import test from the package instead of @playwright/test:

// tests/pricing.spec.ts
import { test, expect, cel } from '@varianz/playwright';

test('payment decline shows an error', async ({ page, varianz }) => {
  await varianz.insert(cel('payment/charge',
    'ChargeResult { txnId: "", declined: true }'));
  await varianz.awaitSyncOrFail(5_000);

  // page already carries x-varianz-id = varianz.sessionId
  await page.goto('/cart');
  await page.click('#checkout-button');

  await expect(page.locator('#error-message')).toContainText('declined');
  await expect(varianz).toEventuallyHaveSample('order-placed', { within: 10_000 });
});

The varianz fixture is test-scoped; a worker-scoped VarianzClient handles the registry connection. The session API (insert, awaitSyncOrFail, assertSample, sessionId) is the same as Vitest’s.

Endpoint

Set the registry endpoint explicitly — via VARIANZ_ENDPOINT in the test environment:

VARIANZ_ENDPOINT=http://localhost:50051 \
VARIANZ_ENABLED=true \
VARIANZ_INSECURE_ALLOW_PLAINTEXT=true \
npx playwright test

What Playwright tests can and can’t reach

The browser only talks to your frontend, but the session header propagates from the frontend through the backend graph — so stages on any downstream VPoint (payments, email, fraud) apply to the browser’s requests. The backend services can be written in any supported language; the frontend itself only needs to forward the header.

Getting started

Core concepts

Guides

SDK reference

Testing

Reference

Endpoint

What Playwright tests can and can’t reach

​Endpoint

​What Playwright tests can and can’t reach

Endpoint

What Playwright tests can and can’t reach