mcp-openvision 0.1.0

MCP server for image analysis using OpenRouter

MCP OpenVision

A simple MCP server for image analysis using OpenRouter. This MCP server allows clients (like Claude Desktop) to analyze images with state-of-the-art vision models.

Features

• Analyze images from local or remote sources using Claude, GPT-4o, or other models
• Extract text from images with language hints
• Compare images and identify differences
• Multiple analysis modes (general, detailed, text extraction, technical, social media)
• Configurable through environment variables

Installation

Option 1: Install from PyPI

pip install mcp-openvision

Option 2: Install from source

# Clone the repository
git clone https://github.com/Nazruden/mcp-openvision.git
cd mcp-openvision

# Install the package
pip install -e .

Option 3: Install with FastMCP

# Install directly for use with Claude
fastmcp install src/mcp_openvision/server.py --name "OpenVision" -e OPENROUTER_API_KEY=your_api_key_here

Requirements

• Python 3.10 or higher
• An OpenRouter API key (get one at openrouter.ai)
• Claude Desktop or another MCP client

Configuration

MCP OpenVision can be configured using environment variables:

• OPENROUTER_API_KEY (required): Your OpenRouter API key
• OPENROUTER_DEFAULT_MODEL (optional): The default vision model to use (defaults to "anthropic/claude-3-sonnet")

Adding to your mcp.json

To use OpenVision with Claude Desktop or other MCP clients, add this to your mcp.json file:

{
  "mcpServers": {
    "openvision": {
      "command": "uvx",
      "args": ["mcp-openvision"],
      "env": {
        "OPENROUTER_API_KEY": "your_openrouter_api_key_here",
        "OPENROUTER_DEFAULT_MODEL": "anthropic/claude-3-sonnet"
      }
    }
  }
}

For more detailed configuration options, see MCP_CONFIG.md.

Example Prompts

• "Analyze this screenshot and tell me what's happening on the webpage"
• "Extract all text from this image"
• "Describe this photo in detail"
• "Is there a dog in this picture?"
• "What kind of chart is this and what data is it showing?"

Development

Running in Development Mode

For development and testing, run the server in development mode:

fastmcp dev src/mcp_openvision/server.py -e OPENROUTER_API_KEY=your_api_key_here

Running Tests

pip install pytest pytest-asyncio
pytest tests/

Available Tools

The MCP server provides these tools:

• analyze_image: General purpose image analysis with various modes
• extract_text_from_image: Specialized tool for extracting text from images
• compare_images: Tool for comparing two images and describing differences

License

MIT