playwright-test-data

name	playwright-test-data
description	Use when creating, seeding, and cleaning test data for Playwright suites, especially under parallel execution. Defines deterministic factories, per-worker namespacing, and teardown discipline.

Playwright Test Data

Bad test data strategy is the root of most "fails in CI only" issues. This skill defines deterministic, parallel-safe data patterns.

Non-Negotiables

1. Every mutable entity created by tests must be uniquely namespaced.
2. Worker-scoped shared data must use testInfo.workerIndex or testInfo.parallelIndex.
3. Data creation and cleanup belong in fixtures, not beforeAll/afterAll.
4. Random data is allowed only when seedable or logged for replay.
5. Tests must not depend on pre-existing production-like shared records.

Factory Pattern

type UserInput = { role?: 'admin' | 'member' };

type TestUser = { email: string; password: string; id?: string };

export function buildUser(input: UserInput, runId: string, laneIndex: number): TestUser {
  const role = input.role ?? 'member';
  return {
    email: `${role}-${runId}-lane${laneIndex}@test.dev`,
    password: 'pw-Strong-1234',
  };
}

Fixture Pattern (Create + Teardown)

import { test as base } from '@playwright/test';

type Fixtures = { testUser: { id: string; email: string; password: string } };

export const test = base.extend<Fixtures>({
  testUser: async ({}, use, testInfo) => {
    const runId = process.env.CI_PIPELINE_ID ?? Date.now().toString();
    const email = `user-${runId}-w${testInfo.workerIndex}@test.dev`;

    const created = await api.createUser({ email, password: 'pw-Strong-1234' });
    await use(created);
    await api.deleteUser(created.id);
  },
});

Determinism Rules

• Freeze time for time-sensitive tests when possible.
• Seed pseudo-random generators and log seed in CI output.
• Prefer explicit IDs from mocks when testing UI flows.
• Use parallelIndex rather than workerIndex when reusing pre-created accounts or cached state after a worker restart.

Anti-Patterns

• Static email like [e-mail] shared across workers.
• Global records created once and mutated by many tests.
• Cleanup only in afterAll, causing cross-test leakage mid-run.
• Test data generated inside specs with ad-hoc helpers.

Cross-References

• Isolation policy: playwright-test-isolation
• Fixture design: playwright-fixtures
• Auth bootstrap data: playwright-auth
• CI sharding interactions: playwright-ci

Quick Quality Checklist

• Every test-created record is uniquely namespaced.
• Worker-scoped fixtures include testInfo.workerIndex or testInfo.parallelIndex.
• Fixtures create and teardown data in the same scope.
• Failures are reproducible from logged identifiers/seeds.
• No test depends on mutable shared baseline records.