Mcp Ocr Server

MCP OCR Server

[](https://github.com/dangvinh/mcp-ocr-server/actions/workflows/ci.yml) [](https://codecov.io/gh/dangvinh/mcp-ocr-server)

MCP server for OCR using native Tesseract (C++), built with Node.js, delivering high-performance OCR and integrable with ChatGPT Desktop.

🚧 Work in Progress 🚧

✨ Features (planned)

• High-performance OCR via native Tesseract (C++)
• Node.js MCP server wrapper for easy integration
• Compatible with ChatGPT Desktop and other MCP clients
• Benchmark vs tesseract.js

📌 Roadmap

• [ ] Step 1: C++ OCR CLI tool
• [ ] Step 2: Node.js MCP server wrapper
• [ ] Step 3: ChatGPT Desktop configuration guide
• [ ] Step 4: Benchmark results
• [ ] Step 5: Demo video

🔧 Tech Stack

• C++ (Tesseract OCR)
• Node.js + TypeScript (@modelcontextprotocol/sdk)
• JSON-RPC 2.0 (MCP standard)

🛠 Installation

1. Install Tesseract OCR

macOS

brew install tesseract
# Optional: install additional languages
brew install tesseract-lang

Linux (Ubuntu/Debian)

sudo apt update
sudo apt install tesseract-ocr libtesseract-dev libleptonica-dev
# Optional: install Vietnamese language
sudo apt install tesseract-ocr-vie

Windows

• Download installer from Tesseract OCR GitHub
• Or using Chocolatey:

choco install tesseract

• Add the installation path to your PATH environment variable.

2. Clone the repository

git clone https://github.com/dangvinh/mcp-ocr-server.git
cd mcp-ocr-server/cpp

3. Build the project with CMake

You can build the C++ OCR engine using the provided npm script. Run:

npm run build-core

This command will create the cpp/build-core directory, configure the project with CMake, and build the static library and CLI tool.

What it builds:

• libmcp_ocr.a static library
• ocr_cli executable in cpp/build-core/bin (or equivalent)

Running tests

# From the build directory
ctest --verbose

• This will run all GoogleTest-based tests.
• Ensure test images or resources exist in cpp/tests or examples/.
• The setup works cross-platform (macOS, Linux, Windows).

4. Build the Node.js addon

The Node.js addon can be built using the provided npm script. Run:

npm run build-addon

This command runs node-gyp inside the cpp/ directory and produces the compiled addon (ocr_addon.node) inside cpp/build/Release/. This addon is required for Node.js integration with the C++ core.

---

🗂 Setup tessdata

The OCR engine requires trained data files to work. Please follow these steps:

1. Create a tessdata folder in the project root:

mkdir tessdata

1. Download the English trained data:

wget https://github.com/tesseract-ocr/tessdata/raw/main/eng.traineddata -P tessdata/

1. For other languages, download the corresponding .traineddata files into tessdata/.

1. Ensure your .env or .env.example has:

TESSDATA_PREFIX=./tessdata
OCR_LANG=eng

---

4. Run OCR CLI

./ocr_cli path/to/image.png

> Ensure the tessdata folder is accessible for language files. The project supports macOS, Linux, and Windows (cross-platform).

Source & license

This open-source MCP server is cataloged on AgentStack and links to its original source — we do not rehost the code.

• Author: dangvinh
• Source: dangvinh/mcp-ocr-server
• License: MIT

Install and usage instructions live in the source repository linked above.