PDF Reader MCP Server

A Model Context Protocol (MCP) server for extracting and processing content from PDF documents. This server provides secure, efficient, and flexible PDF content extraction capabilities following the MCP specification.

Features

• Text Extraction: Extract plain text from PDF documents with formatting preservation
• Metadata Extraction: Extract document metadata (title, author, dates, page count, etc.)
• Page-Level Processing: Extract content from specific pages or page ranges
• PDF Validation: Validate PDF file integrity and readability
• Security-First: Input validation and sandboxed processing
• Type-Safe: Full TypeScript implementation with comprehensive type definitions

Why Choose This MCP Server?

🎯 Specialized PDF Tools

• 4 dedicated tools for different PDF processing needs (text, metadata, pages, validation)
• Granular control - extract specific pages, preserve formatting, or get structured output
• Flexible page ranges - support for "1-5", "1,3,5", or "all" syntax

🛡️ Enterprise-Grade Security

• Directory traversal protection prevents unauthorized file access
• File size limits (configurable up to 100MB by default)
• Processing timeouts prevent resource exhaustion
• Memory usage controls (500MB limit by default)
• No temporary file persistence - secure processing without data leakage

⚡ Production-Ready Architecture

• Robust error handling with standardized MCP error codes (-32602 to -32605)
• Structured logging with Winston for monitoring and debugging
• Comprehensive input validation using Zod schemas
• Type-safe TypeScript implementation with full type definitions
• Concurrent processing support for multiple PDF operations

🔧 Developer Experience

• Easy configuration via environment variables
• Flexible deployment - works with 70+ MCP-compatible clients
• Clear documentation with real-world examples
• Modern tech stack - TypeScript, pdf-parse, pdf-lib
• Test coverage with Vitest for reliability

📊 Performance Optimized

• Efficient PDF processing optimized for text-based documents
• Configurable resource limits to match your infrastructure
• Minimal dependencies for faster startup and lower memory footprint
• Streaming support for large document processing

Installation

npm install
npm run build

Usage

As MCP Server

Start the server:

npm start

Client Configuration

This MCP server can be used with various AI applications and development tools. Below are configuration instructions for the most popular clients:

Claude Desktop

Add this configuration to your Claude Desktop config file:

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

{
  "mcpServers": {
    "pdf-reader": {
      "command": "node",
      "args": ["/path/to/pdf-reader-mcp/dist/index.js"],
      "env": {
        "PDF_MAX_FILE_SIZE": "104857600",
        "LOG_LEVEL": "info"
      }
    }
  }
}

VS Code and VS Code-based Editors

For VS Code, Cursor, Windsurf, and other VS Code-based editors, install an MCP extension:

1. Install the MCP extension from the marketplace
2. Add this configuration to your settings.json:

{
  "mcp.servers": {
    "pdf-reader": {
      "command": "node",
      "args": ["/path/to/pdf-reader-mcp/dist/index.js"]
    }
  }
}

ChatGPT Desktop

For ChatGPT Desktop (available since OpenAI's MCP adoption in March 2025):

1. Go to Settings → Integrations → MCP Servers
2. Add new server with:
   • Name: PDF Reader
   • Command: node /path/to/pdf-reader-mcp/dist/index.js

Claude Code

Option 1: Command Line (Recommended)

Unix/macOS:

# Add the MCP server directly via command line
claude mcp add pdf-reader node /path/to/pdf-reader-mcp/dist/index.js

# With environment variables
claude mcp add pdf-reader -e PDF_MAX_FILE_SIZE=104857600 -e LOG_LEVEL=info -- node /path/to/pdf-reader-mcp/dist/index.js

# Set scope (optional: --scope local|project|user)
claude mcp add --scope project pdf-reader node /path/to/pdf-reader-mcp/dist/index.js

Windows:

rem Add the MCP server directly via command line
claude mcp add pdf-reader node C:\path\to\pdf-reader-mcp\dist\index.js

rem With environment variables
claude mcp add pdf-reader -e PDF_MAX_FILE_SIZE=104857600 -e LOG_LEVEL=info -- node C:\path\to\pdf-reader-mcp\dist\index.js

rem Set scope (optional: --scope local|project|user)
claude mcp add --scope project pdf-reader node C:\path\to\pdf-reader-mcp\dist\index.js

Option 2: Configuration File Configure in your project's .claude/settings.json:

{
  "mcp": {
    "servers": {
      "pdf-reader": {
        "command": "node",
        "args": ["/path/to/pdf-reader-mcp/dist/index.js"]
      }
    }
  }
}

Other Clients

For other MCP-compatible applications (Microsoft Copilot Studio, Replit, Zed, etc.), refer to the official MCP documentation for client-specific configuration instructions.

Available Tools

1. extract_pdf_text

Extract text content from PDF documents.

Parameters:

• file_path (required): Path to the PDF file
• pages (optional): Page range ("1-5", "1,3,5", or "all")
• preserve_formatting (optional): Whether to preserve text formatting
• include_metadata (optional): Whether to include document metadata

2. extract_pdf_metadata

Extract metadata and document information from PDF files.

Parameters:

• file_path (required): Path to the PDF file

3. extract_pdf_pages

Extract content from specific pages or page ranges.

Parameters:

• file_path (required): Path to the PDF file
• page_range (required): Page range to extract
• output_format (optional): "text" or "structured"

4. validate_pdf

Validate PDF file integrity and readability.

Parameters:

• file_path (required): Path to the PDF file

Configuration

Environment variables:

• PDF_MAX_FILE_SIZE: Maximum file size in bytes (default: 104857600 = 100MB)
• PDF_PROCESSING_TIMEOUT: Processing timeout in milliseconds (default: 60000)
• PDF_MAX_MEMORY_USAGE: Maximum memory usage in bytes (default: 524288000 = 500MB)
• LOG_LEVEL: Logging level (default: 'info')

Security

• Input validation for all file paths
• Directory traversal protection
• File size and memory limits
• Processing timeouts
• No temporary file persistence

Error Handling

The server provides comprehensive error handling with specific error codes:

• -32602: Validation errors
• -32603: File access errors
• -32604: Size/resource errors
• -32605: Format errors

Performance

• Supports files up to 100MB
• Memory usage limited to 500MB
• Concurrent processing support
• Optimized for text-based PDFs

License

MIT