AutoUAT Documentation

AutoUAT is a stochastic UAT/UX testing engine. Point it at any URL and it automatically discovers pages, generates test scenarios, executes them in real browsers, and produces structured reports with prioritized issues.

Unlike traditional E2E tools, AutoUAT does not require you to write test scripts. It automatically focuses on the parts of your app most likely to have bugs.

Core Capabilities

• Discovery - Crawls your site, finds routes, forms, auth walls, and interactive elements automatically.
• Scenario generation - Creates role/viewport/flow combinations using adaptive sampling.
• Browser execution - Runs in real Chromium, Firefox, WebKit, or Electron via Playwright with screenshots at every step.
• Heuristic evaluation - 12+ production-ready evaluators score each scenario for layout, accessibility, performance.
• LLM critic panel - Optional multi-agent LLM panel reviews findings with 2-of-3 consensus.
• Triage + dedup - Automatic severity scoring deduplicates and ranks issues.

Interfaces

Quick Start

Get from zero to your first report in under 5 minutes.

GUI Desktop App

Download and install the app, paste a URL into the target field, click Start a Run.

CLI Command Line

# Quick scan with defaults
npm run cli -- run https://example.com

# With options
npm run cli -- run https://example.com --scenarios 30 --depth 5

# Replay a previous run
npm run cli -- replay <runId>

MCP AI Agents

# Zero-config scan from any MCP client
autouat_scan({ url: "https://example.com" })

# Check findings from a run
autouat_findings({ runId: "run-2026-03-09-..." })

How It Works

AutoUAT follows a six-stage pipeline for every run.

Run Setup

The Run Setup screen is where every test begins. Enter a URL, tune parameters, and launch.

Reports

Every run produces multiple report formats to suit different workflows.

Review Studio (Interactive HTML)

The primary report. Screenshot overlays, element inspection, findings panel, and critic panel are all included. State persists in localStorage between sessions.

Standalone Export

A single self-contained HTML file you can email or share - no server required.

Markdown / LLM Export

Two variants:

• report.md - Full findings in Markdown.
• llm-export.md - Compact format optimized for pasting into LLM conversations.

JSON Artifacts

Machine-readable findings for programmatic consumption, CI assertion, or integration with other tooling.

CI/CD Integration

AutoUAT integrates into any CI/CD pipeline via the CLI as a release gate. Use --release-gate to apply stricter thresholds and --exit-code to return a non-zero exit code when the gate fails.

GitHub Actions

name: AutoUAT Release Gate
on: [push]
jobs:
  quality-gate:
    runs-on: ubuntu-latest
    steps:
      - uses: actions/checkout@v4
      - uses: actions/setup-node@v4
        with:
          node-version: 20
      - run: npx playwright install chromium
      - run: npx autouat run --release-gate --baseUrl $STAGING_URL --exit-code

Generic CLI

npx autouat run \
  --release-gate \
  --baseUrl https://staging.example.com \
  --max-scenarios 30 \
  --exit-code

Quality Gates

Quality gates are pass/fail policies that decide whether a build is safe to release.

Policy File

Commit thresholds to your repo at .autouat/policy.json:

{
  "maxNewCritical": 0,
  "maxUnresolvedRegressions": 0,
  "minPassRate": 0.8
}

Suppression Rules

Auto-dismiss known noise with .autouat/suppressions.json. Useful for third-party iframes, known flaky paths, or accepted issues.

Troubleshooting

Credentials & API Keys

Network & Connectivity

Discovery & SPAs