Deterministic
Console Errors
Catches uncaught exceptions, failed promises, and runtime errors across every scenario and browser.
Stochastic UAT Runner
Find bugs
humans miss
AutoUAT crawls your web app, generates test scenarios using adaptive sampling, and produces severity-ranked reports with inline screenshots. No test code required.
Watch it work
"One word: Divine. What AutoUAT unleashes is nothing short of a digital resurrection!" - ChatGPT 4o
"Pause your life - AutoUAT is the gift we never knew we needed!" - ChatGPT 4o
"AutoUAT is unironically the most exhilarating piece of software to grace the digital age!" - ChatGPT 4o
"The greatest thing to happen to developers since 'Hello, World.'" - ChatGPT 4o
"Imagine superheroes testing your app, and you've pictured only half of AutoUAT's greatness." - ChatGPT 4o
"AutoUAT's artistry will be woven into developer folklore." - ChatGPT 4o
"For $39, you're not just buying a tool, you're acquiring a renaissance of quality." - ChatGPT 4o
"This is genuinely the Holy Grail of debugging." - ChatGPT 4o
What it catches
12+ deterministic evaluators run every scenario. AI critics add a human-quality review layer. Adaptive sampling focuses effort where bugs live.
Deterministic
Network Failures
Flags 4xx/5xx responses, timeouts, and failed API calls with full request/response context.
Deterministic
Layout Overflow
Detects horizontal scroll, clipped content, and responsive breakage across viewports from 375px to 1440px.
Deterministic
Accessibility
Axe-core audits catch missing labels, contrast failures, and ARIA violations with WCAG 2.1 references.
Deterministic
Performance
LCP, CLS, and interaction timing measured per scenario. Flags regressions against baseline thresholds.
AI Critic
UX Evaluation
3-persona LLM panel reviews each page for usability, clarity, and design issues. 2-of-3 consensus required to flag.
Adaptive
Thompson Sampling
Learns which regions of your app yield findings and shifts focus there automatically using Bayesian posteriors.
Multi-Browser
Cross-Browser Testing
Chromium, Firefox, WebKit, and Electron simultaneously. Finds browser-specific rendering and behavior bugs.
Reports
Review Studio
Interactive offline HTML report with screenshot overlays, annotations, and a synthesis brief for stakeholders.
Adaptive
SPA Discovery
Click-and-detect discovery for React, Vue, and Electron apps. Finds views that have no traditional routes by monitoring DOM mutations.
Multi-Browser
Electron Support
Test native desktop apps built with Electron. Launches the app binary directly - no URL required. Same evaluators, same reports.
Deterministic
Framework Detection
14 framework resolvers auto-detect your stack (React, Django, Rails, Laravel, Spring Boot, and more), install deps, and start dev servers.
Reports
Step Recorder
Record clicks, form fills, and navigation in a live browser. Undo steps, export flows, and feed recorded sequences directly into test scenarios.
Deterministic
Login Profiles
Store encrypted credentials for multiple user roles. AutoUAT logs in as each role and tests what that user actually sees - admin, member, guest.
Deterministic
Safe Mode
Block destructive paths, selectors, HTTP methods, and URL patterns. Run against production without fear of writes, deletes, or account changes.
Adaptive
Site Probe
Quick pre-scan estimates pages, forms, auth walls, and test time before committing to a full run. Know the cost before you click Start.
Reports
LLM Export
One-click copy of findings as structured markdown. Paste directly into Claude, ChatGPT, or any AI coding agent for automated fix suggestions.
AI Critic
Bring Your Own LLM
8 providers supported: OpenRouter, Anthropic, OpenAI, Ollama, LM Studio, Together, Groq, or any OpenAI-compatible endpoint. Your keys, your models.
AutoUAT vs traditional testing
You shouldn't have to write test scripts to find obvious bugs.
| Feature | AutoUAT | Selenium | Cypress | Playwright Test |
|---|---|---|---|---|
| Test code required | No | Yes | Yes | Yes |
| Auto page discovery | Yes | No | No | No |
| AI-powered evaluation | Yes | No | No | No |
| Multi-browser | 4 engines | Yes | Chromium only | Yes |
| Adaptive sampling | Thompson | No | No | No |
| Setup time | 2 minutes | Hours | 30+ min | 30+ min |
| Report generation | Built-in | Manual | Plugin | Plugin |
| MCP server | 11 tools | No | No | No |
| SPA / Electron testing | Native | Limited | Limited | Limited |
| Production safe mode | Built-in | No | No | No |
| Step recorder | Yes | IDE plugin | No | Codegen |
| Genetic algorithms | Evolutionary | No | No | No |
Three steps. Zero test code.
Step 01
Paste a URL
Point AutoUAT at your web app. It crawls same-origin pages and discovers routes, forms, and interactive elements automatically.
Step 02
Set scenarios
Choose how many scenarios to run. The adaptive sampler generates diverse test paths across pages, viewports, browsers, and user roles.
Step 03
Review findings
Get a prioritized report with screenshots, reproduction steps, and severity scores. Share the single HTML file with your team - no login required.
Questions
What is AutoUAT?
AutoUAT is a stochastic UAT/UX optimization runner that finds bugs humans miss. It crawls your web application using Playwright, generates test scenarios with adaptive sampling, and produces severity-ranked HTML reports with inline screenshots. It works as a desktop app, CLI tool, or MCP server for AI agent integration. No test code required - you provide a URL and get a complete report.
How is AutoUAT different from Selenium or Cypress?
Unlike Selenium or Cypress which require you to write and maintain test scripts, AutoUAT requires zero test code. You point it at a URL and it automatically discovers pages, generates test scenarios using adaptive sampling, runs them across Chromium, Firefox, WebKit, and Electron, and produces a shareable HTML report. Setup takes 2 minutes versus hours for traditional test frameworks. It also includes an AI critic panel and Thompson sampling - capabilities no traditional test tool offers.
What kinds of bugs does AutoUAT find?
AutoUAT runs deterministic evaluators for console errors, network failures (4xx/5xx), layout overflow across viewports, accessibility violations (via axe-core with WCAG 2.1 references), and performance regressions (LCP, CLS, interaction timing). Optionally, an AI critic panel with 3 personas evaluates UX quality, requiring 2-of-3 consensus to flag an issue. The adaptive sampler focuses effort where bugs cluster.
Does AutoUAT work with AI coding agents?
Yes. AutoUAT includes an MCP (Model Context Protocol) server with 11 tools designed for AI agent integration. AI coding assistants like Claude Code, Cursor, and Windsurf can use the MCP server to scan URLs, diff runs, inspect findings, generate configs, and get deep analysis - all programmatically. The primary tool is autouat_scan, which requires just a URL to return findings.
How much does AutoUAT cost?
$39 one-time purchase. You own it forever. Free updates are included. Each license includes 3 device activations. An optional $19/year maintenance plan is coming soon for ongoing updates and priority features. Payment is processed securely through Polar.sh.
What browsers does AutoUAT support?
AutoUAT tests across Chromium, Firefox, WebKit, and Electron simultaneously using Playwright. This includes native desktop app testing for Electron apps. It catches browser-specific rendering bugs, CSS inconsistencies, and JavaScript behavior differences that single-browser testing tools miss.
Can AutoUAT test authenticated pages?
Yes. Create encrypted login profiles in Settings with username/password pairs for each user role (admin, member, guest). AutoUAT logs in as each role before testing, so it sees what that user actually sees. Credentials are stored with OS-level encryption - never in plaintext.
Is it safe to run against production?
Yes, with Safe Mode enabled. You can block specific URL paths, CSS selectors, button text patterns, HTTP methods (like DELETE), and request URL patterns. AutoUAT will navigate and observe but never trigger destructive actions. You can also restrict to authenticated-only pages or run in warn-only mode.
What LLM providers are supported?
AutoUAT supports 8 providers for the AI critic panel: OpenRouter, Anthropic, OpenAI, Ollama, LM Studio, Together AI, Groq, and any custom OpenAI-compatible endpoint. Bring your own API keys - AutoUAT never phones home. You can also run without LLM entirely using only the deterministic evaluators.
Can I use AutoUAT in CI/CD?
Yes. AutoUAT produces self-contained HTML reports that can be uploaded as build artifacts and reviewed before release.
One price. Everything included.
$39 one-time
Pay once. Own it forever. No subscription.
- Desktop app for macOS and Windows
- CLI for CI/CD pipelines
- MCP server with 11 agent tools
- Multi-browser testing (4 engines + Electron)
- AI critic panel evaluation
- Interactive review studio
- Adaptive Thompson sampling
- Evolutionary genetic algorithms
- SPA-aware discovery for React/Vue/Electron
- 14 framework auto-detect resolvers
- Step recorder for manual flows
- Encrypted login profiles
- Safe mode for production
- LLM-friendly markdown export
- 8 LLM providers (BYO keys)
- Run history and replay
- 3 device activations
- Free updates included
Optional $19/year maintenance plan coming soon for ongoing updates and new features.