woladi/macos-vision-mcp

What it does

woladi/macos-vision-mcp is an MCP server that wraps macOS's native Vision API to provide powerful local OCR and image analysis capabilities. It enables optical character recognition on images and PDFs with detailed metadata including reading order paragraphs, bounding boxes, and confidence scores. The tool also supports face detection, barcode recognition, QR code scanning, document corner detection, and image classification—all processed locally on your machine.

How to set up

This is an MCP (Model Context Protocol) server designed for macOS systems. Installation involves cloning the repository from GitHub and configuring it as an MCP server within your AI tool's configuration. Ensure you have macOS with the Vision framework available, then add the server endpoint to your Claude or compatible AI client settings.

Key features

• OCR with line and paragraph IDs and confidence metrics
• PDF text extraction with reading order preservation
• Bounding box coordinates for precise text location
• Face detection and recognition
• Barcode and QR code scanning
• Document corner detection for document scanning
• Image classification capabilities
• Fully local processing—no external API calls required