jerry

AI QA Showcase

A working showcase of what QA engineering looks like with AI in the loop — tests that write themselves from plain English or Gherkin, selectors that heal when the UI changes, an AI agent that navigates your app like a human, and a semantic judge for non-deterministic outputs.

Features

⟳Playwright

Self-Healing Tests

4-strategy fallback chain — data-testid → ARIA → text → CSS. Selectors that recover on their own.

◆Claude

AI Test Generator

Plain English or Gherkin in, executable Playwright TypeScript out. Run it immediately.

◈MCP

Playwright MCP Agent

Natural language drives a real headless browser via Claude's tool-use loop — not a simulation.

◉Quality Gates

Lighthouse Audits

Live Performance, Accessibility, Best Practices, and SEO scored against your thresholds.

⊡Validation

LLM-as-a-Judge

Semantic similarity scoring for AI outputs. Catches meaning changes, ignores cosmetic ones.

jerry · Playwright + Claude + Supabase