Automates post-deploy browser QA: smoke tests, UI interaction checks, visual regression and accessibility audits.
- What it does
- Guides an agent through browser automation of a live or preview page in four phases: smoke testing (console errors, 4xx/5xx requests, screenshots, Core Web Vitals), interaction testing (navigation links, form submission, auth flows, key user journeys), visual regression at three breakpoints, and accessibility checks with axe-core or equivalent. It produces a structured markdown QA report with pass/fail items, flagged issues and screenshots. It also suggests pairing with benchmark, canary-watch and pull-request workflows.
- When to use it
- Use after deploying a feature to staging or preview, when page-wide UI behavior needs verification, before shipping to confirm layout, forms and interactions work, when reviewing frontend pull requests, or for accessibility and responsive testing.
- Requirements
- A browser automation MCP such as claude-in-chrome, Playwright or Puppeteer, plus network access to the target URL. Accessibility checks need axe-core or an equivalent tool. Instructions only; no scripts are shipped.