This workshop is for frontend engineers who already use AI coding tools (Codex, Cursor, Gemini, etc.) and want a reliable verification and investigation workflow for UI changes. You should have light familiarity with Playwright and understand that it is an end-to-end or integration test tool; we will do a focused refresher, but we will not spend half the course on locator basics. If you are responsible for shipping changes into real codebases and want verification that survives refactors and CI, this course is designed for you.

AI coding tools are fast at editing code but blind to whether the UI actually behaves after the change. This course teaches you to use Playwright as a verification and observability layer so an agent (or you) can prove outcomes, detect regressions, and generate high-signal evidence from the running app. You’ll build a complete agent verification harness: deterministic tests that encode acceptance criteria, an evidence pipeline that emits traces, screenshots, logs, and structured failure summaries, and a controlled UI probe tool with guardrails and safe defaults. The workflow is model-agnostic and editor-agnostic, centered on Node.js and TypeScript, and ends with a capstone that runs in CI with artifacts that make failures explainable.