MCP Enabled PDF Reader

by karateboss

5 376 downloads Not rated yet Apache-2.0
GitHub

About

Model Context Protocol (MCP) server to read a single pdf document

Details

License
Apache-2.0

Explore

- Exposes one tool: read_pdf
- Reads a single PDF document per invocation
- No maximum PDF file size (token limit is the limiting factor)
- Tested with Claude Desktop and LibreChat with Ollama
- Configured via claude_desktop_config.json using uvx
- Developed and maintained by Safe Swiss Cloud

Setting up with Highlight

This MCP is not yet compatible with Highlight’s one-click setup. However, you can still use it with Highlight by following these steps:

  1. Download and install Highlight from highlightai.com/download
  2. Navigate to the plugins tab and select "Add Custom Plugin"
  3. Configure the plugin with the settings below
    Plugin Name MCP Enabled PDF Reader
    Command (node, npx, python, etc.)

    Please refer to the README for specific instructions on how to obtain API keys or other required environment variables.

  4. Enable "Start Automatically" if you want the plugin to start when Highlight launches

From the repository

Add the following to your claude_desktop_config.json:

{
    "mcpServers": {
        "mcp-pdf-reader": {
            "command": "uvx",
            "args": [
                "--from",
                "git+https://github.com/karateboss/mcp_pdf_reader@main",
                "mcp_pdf_reader"
            ]
        }
    }
}

Claude Desktop / Cursor

Paste into your MCP client config file to install this server.

{
    "mcpServers": {
        "mcp enabled pdf reader": {
            "mcp-pdf-reader": {
                "command": "uvx",
                "args": [
                    "--from",
                    "git+https://github.com/karateboss/mcp_pdf_reader@main",
                    "mcp_pdf_reader"
                ]
            }
        }
    }
}

McpServers

{
    "mcp-pdf-reader": {
        "command": "uvx",
        "args": [
            "--from",
            "git+https://github.com/karateboss/mcp_pdf_reader@main",
            "mcp_pdf_reader"
        ]
    }
}
Model Context Protocol (MCP) server exposes a tool called read_pdf to read a single pdf document. This has been tested on Claude Desktop and LibreChat with Ollama. Although there is no maximum size to the pdf file that can be read, the number of tokens passed to the model will be the limiting factor.

Installation

Prerequisites

Windows
1. Install Claude Desktop (or another MCP-enabled AI tool) - Download Claude for Desktop - Follow the current installation instructions: Installing Claude Desktop 2. Install Python 3.10 or higher: - Download the latest Python installer from python.org - Run the installer, checking "Add Python to PATH" - Open Command Prompt and verify installation with python --version

3. Install uv:
- Open Command Prompt as Administrator
- Run pip install --user uv
- Verify installation with uv --version

macOS
1. Install Claude Desktop (or another MCP-enabled AI tool) - Download Claude for Desktop - Follow the current installation instructions: Installing Claude Desktop 2. Install Python 3.10 or higher: - Using Homebrew: brew install python - Verify installation with python3 --version

3. Install uv:
- Using Homebrew: brew install uv
- Alternatively: pip3 install --user uv
- Verify installation with uv --version

Configuration

Add the following to your claude_desktop_config.json:

{
    "mcpServers": {
        "mcp-pdf-reader": {
            "command": "uvx",
            "args": [
                "--from",
                "git+https://github.com/karateboss/mcp_pdf_reader@main",
                "mcp_pdf_reader"
            ]
        }
    }
}

Attribution

This software package implements the ability to read a pdf file into a MCP enabled framework and is developed by Safe Swiss Cloud.

Contributing

We welcome contributions to improve these tools. Please submit issues and pull requests through our repository.

Support

For questions and support:
1. Check our documentation
2. Submit an issue in our repository

No reviews yet — be the first

Sign in to leave a review

Use Google, GitHub, or an email account so ratings stay tied to real people.

Email sign in

No reviews posted yet.