bruniai

AI-powered visual regression testing tool - core comparison library.

This package provides the core comparison functionality for visual regression testing. It can be used standalone or as a dependency for building custom integrations (e.g., MCP servers, Next.js apps, etc.).

Installation

npm install bruniai

Usage

import { compareImageToUrl, compareUrls } from "bruniai";

const result = await compareUrls({
  baseUrl: "https://example.com",
  previewUrl: "https://preview.example.com",
  page: "/contact",
});

console.log(result.status); // "pass" | "fail" | "warning"
console.log(result.visual_analysis);
console.log(result.images.base_screenshot);

const imageResult = await compareImageToUrl({
  baseImageSource: "https://example.com/design.png",
  previewUrl: "https://preview.example.com",
  page: "/contact",
});

console.log(imageResult.status);

API

compareUrls(input: CompareUrlsInput): Promise<CompareUrlsOutput>

Performs a visual comparison between two URLs.

Parameters:

• baseUrl (string): Base/reference URL to compare against
• previewUrl (string): Preview/changed URL to analyze
• page (string, optional): Page path to compare (default: "/")
• prNumber (string, optional): PR number for metadata
• repository (string, optional): Repository name for metadata

Returns:

• status: Overall comparison status ("pass" | "fail" | "warning" | "none")
• visual_analysis: Detailed visual analysis result from AI
• sections_analysis: Formatted sections analysis text
• images: Object containing paths to generated screenshots and diff images

compareImageToUrl(input: CompareImageToUrlInput): Promise<CompareImageToUrlOutput>

Performs a visual comparison between a base image source and a preview URL.

Parameters:

• baseImageSource (string): Base/reference image source as an HTTP(S) image URL or data:image/...
• previewUrl (string): Preview/changed webpage URL to analyze
• page (string, optional): Page path to compare on the preview URL (default: "/")
• prNumber (string, optional): PR number for metadata
• repository (string, optional): Repository name for metadata

Returns:

• status: Overall comparison status ("pass" | "fail" | "warning" | "none")
• visual_analysis: Detailed visual analysis result from AI
• sections_analysis: Formatted sections analysis text
• images: Object containing paths to normalized screenshots and diff images

Requirements

• Node.js 18+
• OpenAI API key (set as OPENAI_API_KEY environment variable)

Development

Building

First, build the parent package to generate the dist files:

cd ../..
npm run build

Then build this package:

npm run build

License

MIT