What problem does it solve? Writing Playwright tests by hand often produces brittle specs with guessed CSS selectors, fixed sleeps, and assertions that pass without verifying real behavior. This Skill enforces a disciplined authoring workflow that derives stable locators from the live accessibility tree and requires every test to verify actual application state before claiming success. ## Core Features & Use Cases - MCP-Derived Locators: Drives a real browser through the Playwright MCP (browser_snapshot, browser_generate_locator) to produce stable role/label/text locators instead of brittle CSS selectors. - Two-Layer Assertion Rules: Enforces paired state-layer and visible-layer assertions for every stateful action, banning false-green patterns like waitForTimeout, networkidle waits, and OR'd success branches. - Run-and-Paste Verification: Requires executing the spec against a live server and pasting the pass/fail summary before any completion claim. - Use Case: A developer asks to write e2e tests for a checkout page; the Skill navigates the live app via the Playwright MCP, derives role-based locators for the form fields, authors a spec asserting both the URL change and rendered confirmation, runs it, and reports the result. ## Quick Start Ask the AI to generate Playwright e2e tests for a specific page or user flow in this project.
