juergenkoller-software/freezetext-mcp
🔌 MCP Serverjuergenkoller-software
An MCP server that enables AI agents to perform OCR and screen capture on macOS using FreezeText.
The freezetext-mcp server acts as a powerful bridge between Large Language Models and the macOS desktop environment. It leverages the native Apple Vision framework via the FreezeText app to perform high-accuracy Optical Character Recognition (OCR) on any screen region.
Key features include the ability to capture static images, process base64 encoded image data, and extract text from complex UI elements like protected PDFs, video frames, or temporary popups that standard scrapers often miss. The server exposes 12 distinct tools, allowing agents to trigger screen freezes, perform targeted OCR, and query a local searchable capture history. This integration is particularly useful for automation workflows where an agent needs to read non-selectable text or monitor dynamic application states in real-time. By utilizing the Model Context Protocol (MCP), it ensures a standardized, secure, and efficient communication channel between the AI assistant and the local macOS system.
💡Highlights
- ├─12 specialized OCR tools
- ├─Native Apple Vision integration
- └─Supports video and protected PDFs
🎯For
- ├─AI Automation Developers
- └─macOS Power Users