ux-testing

Let vision-capable AI act as demanding customers and catch product problems before launch.

A Physics-Informed UI/UX Audit Engine. Uses Computer Vision (OpenCV Saliency/Entropy) and Gemini 3.0 Multimodal AI to mathematically quantify design quality, generate heatmaps, and prescribe actionable CSS fixes.

Agent Skills for first-user QA: isolated browser personas, evidence, bounded fixes, and fresh verification.

AI test users that run your site's real flows (signup, dashboard, checkout) in your own browser and report where they got stuck, plus SEO and passive security checks. LangGraph persona agents on FastAPI, a Chrome MV3 extension (WXT + React) that snapshots and redacts the DOM, and a React 19 + Vite + Tailwind v4 dashboard on Supabase Postgres.

An autonomous browser QA agent that explores a running web application, discovers new pages and workflows, fuzzes inputs, detects UI failures, and builds a navigation graph dynamically using Playwright + FastAPI + local LLMs through Ollama.

Persona-based Agentic Testing Testing Platform

Autonomous AI visual testing agent — multi-browser crawling + vision LLM analysis.

Agent skills that show proof or say nothing. Evidence-first audits for Claude Code and Codex.

Autonomous agentic framework that drives a real Chromium browser via a Reason-Observe-Act loop, detects WCAG accessibility violations and UX friction in any web app from a plain-English goal, and delivers a full visual audit report — zero test scripts or code changes needed.

The Sentient Mystery Shopper: an autonomous AI agent that tests signup flows like a real user, scores friction, diagnoses root causes, and proposes fixes.

Persona-driven UX Friction Simulation Engine — LLM-powered cognitive load scoring and conversion friction detection

Simulated customer panel of 100+ diverse personas for Claude Code: reviews a website or concept and returns a prioritised, evidence-backed fix list.

A powerful Flow Tester MCP for parameterized UI/UX testing with enhanced logging and MCP-native artifacts. Built with pure Playwright automation and zero external LLM dependencies.

AI website stress testing with browser personas, built at HackIllinois 2026.

Repair-oriented UX E2E testing for ChatGPT Web with real Chromium, source evidence, patches and regression verification.

Blind app testing with first-time-user agents, semantic coverage, state ledgers, and evidence-backed product improvement lists.

Multi-agent beta testing framework where LLaVA vision agents form committees to test web apps. Agents analyze screenshots, discuss actions, vote on consensus, and catch bugs + security vulnerabilities. Playwright browser automation, Streamlit dashboard, FastAPI backend.

Software-Fabrik für smlfg-CLIs und MCP-Server: Workflow, UX-Gate, Health-Check, Pipeline (iterativ)