Skip to content
Finalist

nan

GenUI Test Harness AI-generated UIs look great in screenshots.

Code Video Cerebral Valley

Core trick

Agent crawls a deployed URL, infers user flows from screenshots, generates end-to-end browser tests, and runs them in a headless browser.

Tags

dev toolsagentsvision

Description

GenUI Test Harness AI-generated UIs look great in screenshots. But do they actually work? GenUI Test Harness is an autonomous QA agent that crawls any deployed web app, uses Gemini to infer what the intended user flows should be, auto-generates end-to-end tests, executes them in a real browser, and delivers a live-streamed test report — complete with annotated browser replays, visual diffs, and actionable fix suggestions. No test code written. No config files. Just paste a URL and watch the agent break your app before your users do.
nnaka/ui-tester★ 00 forksTypeScriptCreated Mar 21, 2026 · last push Mar 21, 2026

README excerpt

# 🧪 GenUI Test Harness

**Paste a URL. Watch an AI agent find every bug.**

Every AI code generator can *build* a UI. None of them can tell you if it actually works. GenUI Test Harness closes that gap. It's a multi-modal agent that looks at your generated site the way a QA engineer would — visually inspecting the UI, inferring what users should be able to do, writing real tests, and executing them live. You see every click, every form fill, every failure — streamed in real time with element highlighting and before/after visual diffs. It's the missing feedback loop between "vibe coding" and shipping.

Zero config. Zero test code. Just results.

---

## 🎬 How It Works

```
Paste a URL → Agent crawls the site → Gemini infers user flows →
Tests auto-generate → Headless browser executes them live →
Streamed results with replays, diffs, and fix suggestions
```

1. **You provide** a URL (and optionally the original prompt that generated the site)
2. **The agent crawls** the page — capturing screenshots, DOM structure, every interactive element, console errors, and page metadata
3. **Gemini analyzes** the UI structure and screenshots to infer the 3–8 most important user flows the app should support
4. **A second Gemini pass** refines the test plan against the actual screenshot, catching hallucinated selectors and navigation issues
5. **A headless browser** executes each test step-by-step — clicking buttons, filling forms, verifying state changes — with real-time streaming to your das

Team