# juergenkoller-software/freezetext-mcp [Health: Active]

**Category:** 🖥️ OS Automation  
**Repository:** https://github.com/juergenkoller-software/freezetext-mcp  
**GitHub Stars:** 1  
**Views:** 3  
**Installs:** 0  
**Upvotes:** 0  
**Directory Page:** https://allmcps.com/mcp/juergenkoller-software-freezetext-mcp

## Description
MCP server for FreezeText — OCR anything on your Mac screen. Freeze the screen and extract text via Apple Vision (videos, popups, protected PDFs), OCR a region or base64 image, and manage a searchable capture history. 12 tools.

## Claude Desktop Quick Installation
Heuristic fallback — verify the package name and runner against the repository README before running it. Uses `npx` (confidence: low):

```json
"mcpServers": {
  "freezetext-mcp": {
    "command": "npx",
    "args": ["-y","juergenkoller-software-freezetext-mcp"],
    "env": {
      "FREEZETEXT_API_PORT": "",
      "FREEZETEXT_API_TOKEN": ""
    }
  }
}
```

**Requires environment variables:** `FREEZETEXT_API_PORT`, `FREEZETEXT_API_TOKEN` — the values above are empty placeholders; fill in real credentials before running (see the repository for what each one is for).

## Documentation

## What juergenkoller-software/freezetext-mcp MCP server does

The juergenkoller-software/freezetext-mcp MCP server gives Claude, Cursor, and other MCP clients access to FreezeText's macOS OCR functions. FreezeText is a separate native application that must be installed, running, and configured with its HTTP API enabled.

The server supports three capture paths: freezing the current screen and extracting its text, OCRing a specified screen rectangle, and processing an image supplied as base64 data. These options cover visible interface content as well as screenshots or images already available to an agent.

It also exposes capture-history operations. Agents can list records, search their text, retrieve a particular entry, add or delete entries, clear the history, and export it as JSON or CSV. Two additional tools read and change the configured OCR languages.

## How it works

An MCP client communicates with the juergenkoller-software/freezetext-mcp MCP server over JSON-RPC stdio. The server translates tool calls into requests to FreezeText's local HTTP API, normally on port 9876. FreezeText performs screen capture, Apple Vision OCR, and history storage; those functions are not implemented in the MCP process itself.

The OCR workflow is local to the Mac. The supplied material states that OCR uses Apple Vision Framework and that no cloud service or data transmission is involved. If an API token is configured in FreezeText, the server can send it as a Bearer token when calling the local API.

## Setup and configuration

The juergenkoller-software/freezetext-mcp MCP server targets macOS 14 Sonoma or later. Install and run the FreezeText app, enable its HTTP API in Settings, and then either use a pre-built release binary or build the MCP server from source with Swift 5.9 or later, available through Xcode 15 or newer.

The source-build commands are:

```bash
git clone https://github.com/juergenkoller-software/freezetext-mcp.git
cd freezetext-mcp
swift build -c release
```

The resulting binary is `.build/release/FreezeTextMCP`. Configure that executable as a stdio MCP server in the client. The server accepts these environment variables:

- `FREEZETEXT_API_PORT`: FreezeText API port; defaults to `9876`.
- `FREEZETEXT_API_TOKEN`: optional token used when a token has been set in FreezeText Settings.

Claude Desktop uses its MCP server configuration file, while Claude Code can register the executable with `claude mcp add`. Cursor and other MCP clients use the same stdio arrangement.

## Tools and capabilities

The twelve tools are grouped as follows:

- Capture and OCR: `capture_screen`, `capture_region`, and `ocr_image`.
- History: `list_history`, `search_history`, `get_history_entry`, `add_history`, `delete_history_entry`, `clear_history`, and `export_history`.
- OCR configuration: `get_ocr_languages` and `set_ocr_languages`.

Together, these tools let an agent extract text from transient or difficult-to-copy content, inspect prior captures, maintain history records, and choose OCR languages supported by the FreezeText app.

## Limitations and notes

This server depends on the FreezeText application; installing only the MCP binary is not sufficient. The app must be running and its HTTP API must be enabled. The project is intended for macOS 14 or later and does not describe support for other operating systems.

The MCP server is MIT-licensed and open source, while the FreezeText app is described separately as free and proprietary. The documentation does not identify a package-manager distribution or a hosted remote endpoint; it describes a locally built or downloaded executable communicating with a local app.

_Full upstream README: https://allmcps.com/mcp/juergenkoller-software-freezetext-mcp/readme_