PDF Agent

Provides metadata, text, image extraction and targeted search for PDFs via MCP for efficient AI-assisted document analysis.

typescript

12

GitHub Stars

typescript

Language

5 months ago

First Indexed

2 months ago

Catalog Refreshed

Documentation & install

Readme and setup notes from the catalogue, plus a client-ready config you can copy for your MCP host.

Installation

Add the following to your MCP client configuration file.

Configuration

View docs

{
  "mcpServers": {
    "vlad-ds-pdf-agent-mcp": {
      "command": "node",
      "args": [
        "PATH_TO_REPO/server/index.js"
      ]
    }
  }
}

You can use the PDF Agent MCP to let AI tools read and analyze PDFs selectively. It extracts metadata, text, images, and outlines, and enables targeted queries that avoid loading whole documents into memory. This helps you work with large PDFs efficiently and stay within token limits while preserving relevant context.

How to use

To use the PDF Agent MCP with your client, provide either a local PDF path or a public PDF URL. The agent will first read metadata and the document outline, then selectively open only the sections that matter to your query. This keeps processing scoped and efficient, avoiding full-document dumps.

How to install

Prerequisites: you must have Node.js LTS installed on your system. Get it from the Node.js website and install the recommended LTS version.

Option 1: DXT Package (Recommended) – quick install flow.

Ensure you completed the required configuration steps below.
Download the latest pdf-agent-mcp.dxt file from releases.
Double-click the .dxt file to install it in Claude Desktop.

Option 2: Manual Installation – full source flow.

Ensure you completed the required configuration steps below.
Clone the repository.
Build the project.

npm install && npm run build

Locate Claude Desktop config file.

macOS: ~/Library/Application Support/Claude/claude_desktop_config.json

Windows: %APPDATA%\Claude\claude_desktop_config.json

Add the MCP server configuration to the config file.

{
  "mcpServers": {
    "pdf-agent": {
      "command": "node",
      "args": [
        "PATH_TO_REPO/server/index.js"
      ]
    }
  }
}

Troubleshooting and logs

If you have trouble loading the extension, verify Node.js is installed and that the built-in Node.js option is disabled in the application settings. Then restart the application and check the logs for details.

Log locations:

macOS: ~/Library/Logs/Claude/mcp-server-PDF Agent MCP.log

Windows: %LOCALAPPDATA%\Claude\Logs\mcp-server-PDF Agent MCP.log

Viewing logs

# macOS
open "$HOME/Library/Logs/Claude/mcp-server-PDF Agent MCP.log"

# macOS (live tail)
tail -f "$HOME/Library/Logs/Claude/mcp-server-PDF Agent MCP.log"

# macOS (clear)
rm "$HOME/Library/Logs/Claude/mcp-server-PDF Agent MCP.log"

# Windows
notepad "%LOCALAPPDATA%\Claude\Logs\mcp-server-PDF Agent MCP.log"

# Windows (clear)
del "%LOCALAPPDATA%\Claude\Logs\mcp-server-PDF Agent MCP.log"

Configuration and logs

This MCP server exposes tools to extract and analyze PDF content. You can access metadata, outline, text, images, and pattern-based searches to focus your AI-assisted reading on relevant sections without loading the entire document.

Notes and usage tips

Do not drag PDFs directly into the chat. Instead, provide the absolute file path or a direct URL to trigger selective processing.

Security and data handling

Operate with local copies when possible to minimize data transfer. Use URL or path-based inputs to ensure predictable processing scope.

Available tools

get_pdf_metadata

Retrieve PDF properties, size, creation date, and basic info.

get_pdf_outline

Extract bookmarks and document outline to understand structure and navigation.

get_pdf_images

Extract images from PDF pages for visual content analysis.

search_pdf

Perform pattern/regex searches over PDF content with contextual results.