What problem does it solve? Manual exploratory QA of a web app is slow and inconsistent, and teams often ship features without verifying real user flows, console errors, broken links, or mobile and RTL behavior. This Skill runs a structured, report-only browser QA pass that documents every issue with evidence and a health score, without touching source code or fixing anything. ## Core Features & Use Cases - Exploratory browser testing: Clicks through every screen, form, modal, and state using Playwright MCP, on desktop (1280x800) and mobile (390x844) viewports, with console and network error checks after every interaction. - Structured issue reporting: Appends each verified issue with severity, category, repro steps, and mandatory screenshots, then computes a weighted 0-100 health score across Console, Links, Functional, UX, Accessibility, Visual, Performance, and Content. - Diff-aware and regression modes: Maps changed files on a feature branch to screens for prioritized testing, and supports --regression runs against a saved baseline.json to mark issues as new, persisting, or resolved. - Use Case: Before presenting a feature as done, run the skill against http://localhost:3000 to get a health score, the top issues with screenshots, and a baseline for future regression comparisons. ## Quick Start Ask the agent to run a QA report on your running app, for example: run a QA report on http://localhost:3000 and focus on the checkout flow.
