AI-powered testing for web and mobile apps. An AI agent explores your app like a real user and reports bugs with reproduction steps, screenshots, and video.
What it finds
- Visual bugs (layout issues, overflow, contrast problems)
- Logical bugs (broken validation, stale state, wrong behavior)
- UX issues (confusing flows, missing feedback)
- Accessibility problems
Each issue includes severity, category, confidence score, and step-by-step reproduction instructions.
Quick start
- Install the extension
- On first activation, the extension automatically:
- Installs the Agentiqa CLI
- Opens browser for authentication
- Installs Chromium for headless testing
- Detects ffmpeg for video recording
- Cmd+Shift+P → "Agentiqa: Test URL" → paste your URL
- Results appear in the terminal with issues, screenshots, and video
Usage
Command Palette
- Agentiqa: Test URL — enter a URL to test, runs explore in terminal
- Agentiqa: Setup — manually re-run CLI install, auth, and browser setup
Copilot Chat
Ask Copilot to test your app:
"test https://example.com/pricing for bugs"
Copilot will run agentiqa explore in the terminal and present the results.
MCP Tools (Agent Mode)
The extension registers an MCP server with these tools:
| Tool | Description |
|---|---|
startExplore |
Start an AI testing agent against a URL. Returns jobId immediately. |
checkExploreStatus |
Poll for progress and results of a running explore. |
startRun |
Start a test plan run against a URL. Returns jobId. |
checkRunStatus |
Poll for progress and results of a running test plan. |
getArtifacts |
List screenshots and video from a completed run. |
Artifacts
Each test run produces:
- Screenshots — one per agent action (always saved)
- Video — MP4 recording of the full session (requires ffmpeg)
Install ffmpeg for video recording:
brew install ffmpeg
Without ffmpeg, tests still work — screenshots are saved as fallback.
Requirements
- VS Code 1.99+
- Node.js 18+
- Internet connection (for CLI install and testing)