PDF Document MCP Server

PDF Document MCP Server

By MEOK AI Labs | meok.ai

Document processing toolkit for AI agents. Extract text, convert to Markdown, merge PDFs, extract tables, and summarize documents -- all locally with no external API dependencies.

Tools

Installation

pip install mcp PyMuPDF

No external services, API keys, or cloud dependencies required. All processing happens locally.

Usage

Run the server

python server.py

Claude Desktop config

{
  "mcpServers": {
    "pdf-document": {
      "command": "python",
      "args": ["/path/to/pdf-document-mcp/server.py"]
    }
  }
}

Example calls

Extract text from a PDF:

Tool: extract_text_from_pdf
Input: {"file_path": "/Users/me/documents/report.pdf"}
Output: {"total_pages": 12, "total_characters": 48320, "pages": [{"page": 1, "text": "...", "char_count": 4210}, ...]}

Extract specific pages only:

Tool: extract_text_from_pdf
Input: {"file_path": "/Users/me/documents/report.pdf", "pages": [1, 3, 5]}
Output: {"extracted_pages": 3, "pages": [...]}

Convert PDF to Markdown:

Tool: convert_pdf_to_markdown
Input: {"file_path": "/Users/me/documents/whitepaper.pdf"}
Output: {"markdown": "# whitepaper.pdf\n\n---\n## Page 1\n\n### Introduction\n\nThis paper presents...", "char_count": 15200}

Extract tables:

Tool: extract_tables
Input: {"file_path": "/Users/me/documents/financials.pdf", "page_num": 3}
Output: {"tables_found": 2, "tables": [{"header": ["Quarter", "Revenue", "Profit"], "rows": [["Q1", "$1.2M", "$400K"], ...]}]}

Merge multiple PDFs:

Tool: merge_pdfs
Input: {"file_paths": ["/tmp/part1.pdf", "/tmp/part2.pdf", "/tmp/part3.pdf"], "output_path": "/tmp/combined.pdf"}
Output: {"output": "/tmp/combined.pdf", "total_pages": 28}

Summarize a document:

Tool: summarize_document
Input: {"file_path": "/Users/me/documents/thesis.pdf"}
Output: {"statistics": {"total_pages": 45, "total_words": 12500, "total_images": 8}, "headings": [...], "metadata": {"author": "...", "title": "..."}}

Pricing

License

MIT