MCP PDF Reader Server

A Model Context Protocol (MCP) server that enables AI assistants like Claude, Windsurf, and other MCP-compatible tools to read and analyze PDF files.

Features

• 📄 Read PDF files from local paths or URLs
• 🔍 Search for specific terms within PDFs
• 📊 Get PDF metadata (page count, author, title, etc.)
• 📖 Read specific pages or page ranges
• 🌐 Download and analyze PDFs from the web
• 🤖 Seamless integration with AI assistants via MCP

Installation

Using npm

npm install -g @dev.saqibaziz/mcp-pdf-reader

Using npx (no installation required)

npx @dev.saqibaziz/mcp-pdf-reader

Configuration

For Claude Desktop

Add this to your Claude Desktop configuration file:

macOS: ~/Library/Application Support/Claude/claude_desktop_config.json Windows: %APPDATA%\Claude\claude_desktop_config.json

{
  "mcpServers": {
    "pdf-reader": {
      "command": "npx",
      "args": ["-y", "@dev.saqibaziz/mcp-pdf-reader"]
    }
  }
}

For Windsurf

Option 1: Via Plugin Store (Coming Soon)

• Open Windsurf
• Click the Plugins icon in Cascade panel
• Search for "PDF Reader"
• Click Install

Option 2: Manual Configuration

Edit your MCP configuration file:

macOS: ~/Library/Application Support/Windsurf/User/globalStorage/codeium.codeium/mcp_config.json Windows: %APPDATA%\Windsurf\User\globalStorage\codeium.codeium\mcp_config.json Linux: ~/.config/Windsurf/User/globalStorage/codeium.codeium/mcp_config.json

{
  "mcpServers": {
    "pdf-reader": {
      "command": "npx",
      "args": ["-y", "@dev.saqibaziz/mcp-pdf-reader"]
    }
  }
}

After adding, click the refresh button in the Plugins panel.

For Cline

Add this to your MCP settings file:

{
  "mcpServers": {
    "pdf-reader": {
      "command": "npx",
      "args": ["-y", "@dev.saqibaziz/mcp-pdf-reader"]
    }
  }
}

Available Tools

The server provides the following tools to AI assistants:

read_pdf_pages

Read specific pages from a PDF file.

Parameters:

• pdf_path (string): Path to the PDF file
• pages (array): List of page numbers to read

read_pdf_page_range

Read a range of pages from a PDF file.

Parameters:

• pdf_path (string): Path to the PDF file
• start_page (number): First page to read (inclusive)
• end_page (number): Last page to read (inclusive)

search_pdf

Search for terms in a PDF file and return pages containing them.

Parameters:

• pdf_path (string): Path to the PDF file
• terms (string or array): Search term(s)

get_pdf_metadata

Get metadata from a PDF file (page count, author, title, etc.).

Parameters:

• pdf_path (string): Path to the PDF file

Usage Examples

Once configured, you can ask your AI assistant:

Local Files:

• "Read pages 1-5 from document.pdf"
• "Search for 'machine learning' in research.pdf"
• "What's the metadata of report.pdf?"
• "Read page 10 from presentation.pdf"

URLs:

• "Read this PDF: https://example.com/research-paper.pdf"
• "Search for 'climate change' in https://example.com/report.pdf"
• "Get metadata from https://arxiv.org/pdf/2301.00001.pdf"
• "Read pages 1-3 from https://example.com/whitepaper.pdf"

Development

Prerequisites

• Node.js >= 18.0.0
• npm or yarn

Setup

# Clone the repository
git clone https://github.com/Saqib-Aziz007/mcp-pdf-reader.git
cd mcp-pdf-reader

# Install dependencies
npm install

# Build the project
npm run build

# Run in development mode
npm run dev

Scripts

• npm run build - Compile TypeScript to JavaScript
• npm run dev - Watch mode for development
• npm start - Run the compiled server
• npm run prepare - Pre-install build hook

Technical Details

This MCP server is built with:

• @modelcontextprotocol/sdk - MCP SDK for TypeScript
• pdf-parse - PDF parsing library
• TypeScript for type safety

Contributing

Contributions are welcome! Please feel free to submit a Pull Request.

License

MIT License - see LICENSE file for details

Author

Muhammad Saqib Aziz

• Email: saqib.aziz1000@gmail.com
• GitHub: @Saqib-Aziz007
• npm: @saqibaziz

Support

If you encounter any issues or have questions:

• Open an issue on GitHub
• Check the MCP documentation

Acknowledgments

Built with the Model Context Protocol by Anthropic.