MCP Image Extractor

Extract relevant images from product and article pages

Test Image Extraction

Enter one or more URLs to extract relevant images from product or article pages.

Get your API key from Google AI Studio. Your key is stored locally in your browser.

What it does

This MCP server extracts only the relevant images from product or article pages, automatically filtering out:

  • Logos and icons
  • Advertisements and tracking pixels
  • Navigation elements
  • Related/recommended products
  • Footer and header images

Returns deduplicated, largest variant images with AI-generated captions in multiple languages.

Integration Instructions

How to Connect

This MCP server uses simple API key authentication. You need a Gemini API key from Google.

đź“‹ What You'll Need:

Server Details:

  • MCP Server URL: https://image.kapler.cz/mcp?api_key=your-gemini-api-key-here
  • ⚠️ Replace your-gemini-api-key-here with your actual Gemini API key

Tool: extract_relevant_images

Input Parameters

{
  "urls": ["https://example.com/product"],  // 1-10 URLs
  "mode": "auto",                           // auto | product | article
  "maxImagesPerSource": 20,                 // 1-50
  "includeSizes": true,                     // probe dimensions
  "languages": ["en", "cs"],                // caption languages
  "api_key": "optional-gemini-key"          // if not in header
}

Output Example

{
  "version": "1.0.0",
  "generatedAt": "2024-01-15T10:30:00Z",
  "sources": [{
    "inputUrl": "https://example.com/product",
    "finalUrl": "https://example.com/product",
    "pageType": "product",
    "mainImage": "https://cdn.example.com/img1.jpg",
    "images": [{
      "url": "https://cdn.example.com/img1.jpg",
      "isMain": true,
      "width": 1200,
      "height": 800,
      "caption": {
        "cs": "Hlavní produktový obrázek",
        "en": "Main product image"
      },
      "confidence": 0.95
    }],
    "warnings": []
  }],
  "errors": []
}

Limits & Security

  • Maximum 10 URLs per request
  • Maximum 50 images per source
  • 30-second timeout per URL fetch
  • SSRF protection (blocks private IPs)
  • Origin validation for remote connections
  • Authentication via Gemini API key