NameetP/pdfmux

What it does

pdfmux is a PDF extraction router that automatically classifies each page in a PDF document (digital, scanned, or table-based) and routes it to the most appropriate processing backend. It uses PyMuPDF, Docling, OCR, and optional LLM fallback to extract content with maximum accuracy. The tool includes per-page confidence scoring that flags low-quality extractions and automatically re-processes them, preventing silent failures in RAG (Retrieval-Augmented Generation) systems.

How to set up

Installation is straightforward with zero configuration required. Simply run pip install pdfmux to install the package. The tool comes with a built-in MCP (Model Context Protocol) server, making it easy to integrate into AI workflows and applications without additional setup steps.

Key features

• Automatic page classification (digital, scanned, tables)
• Intelligent routing to optimal extraction backends
• Per-page confidence scoring with quality flagging
• Automatic re-extraction of low-quality pages
• Built-in MCP server for easy integration
• OCR support for scanned documents
• Zero-configuration installation
• Prevents silent RAG failures through quality monitoring