dongokovacs/playwright_ts

This repository is a demonstration portfolio for Playwright-based test automation in TypeScript. The author built it to show how they would architect a test suite at a senior level, not just how individual tests look. The emphasis is on structure, patterns, and decisions rather than coverage over a specific application. The test suite is split into four areas: pure API tests against a public Conduit REST API, UI tests against a QA Playground site, hybrid tests that write data through the API and verify results through the real browser, and AI-assisted tests that use a language model for semantic checks on text that has no fixed expected value. The AI integration goes through an interface backed by OpenRouter, so nothing in the test files calls an AI API directly. The framework patterns on display include fixture-based dependency injection (all shared setup flows through Playwright fixtures, not global state), a fluent API client with request and response schemas, custom assertion matchers that automatically attach failure context, and a self-healing locator that logs when it falls back to an alternative selector. Page Objects model a single screen, a separate Flow layer assembles them into a named business scenario. Three CI pipelines are included: a fast smoke gate that runs on every pull request, a full test suite that publishes results as a GitHub check, and a nightly run with flaky-test detection. No credentials are required to run the tests locally. A test user account is created per worker automatically. You can also run the entire suite in Docker, which skips the local Node and browser setup and uses the same pinned environment as CI. A separate ARCHITECTURE.md file explains the reasoning behind the layout choices.