Pdf Reader

pdf-reader-mcp

An MCP server that can extract text from PDFs and images. It supports:

• Native text extraction from PDFs (embedded text)
• OCR for scanned PDFs and images using Tesseract

Requirements

• Node.js >= 18.18
• For best OCR on PDFs, install Poppler (for pdftoppm) and Tesseract (CLI):

macOS

brew install poppler tesseract

Windows

1. Tesseract: Download and install from GitHub releases or use chocolatey:

   choco install tesseract
   

2. Poppler: Download from poppler-windows and add to PATH, or use chocolatey:

   choco install poppler
   

Linux (Ubuntu/Debian)

sudo apt-get install poppler-utils tesseract-ocr

Note: Poppler is optional; without it we fall back to slower whole-file OCR. Tesseract CLI is preferred for OCR; if it is not installed, the server falls back to tesseract.js (which downloads language data on first run).

Install

npm install

Build

npm run build

Run (stdio)

npm start

This server uses stdio transport per MCP. Configure your MCP client with the command node dist/index.js in this folder.

Example MCP client configuration (JSON):

{
	"mcpServers": {
		"pdf-reader-mcp": {
			"command": "node",
			"args": ["dist/index.js"],
			"env": { }
		}
	}
}

Tool: extract_text_from_pdf

Inputs (one of):

• path: Path to a local PDF or image file
• base64: Base64-encoded content (optionally with data URL prefix)

Options:

• ocr (boolean, default false): Force using OCR. When false, the server tries native extraction then falls back to OCR if needed.

Output: The extracted text as a single text item.