docling-project/docling-mcp

What it solves

Docling MCP allows AI agents and LLM clients to interact with documents by providing a standardized way to convert, process, and generate structured documents. It bridges the gap between raw PDFs and agentic workflows by turning document processing capabilities into tools that an LLM can call.

How it works

It implements the Model Context Protocol (MCP), which allows it to act as a server that exposes specific tools to an AI client (like Claude for Desktop or LM Studio). It uses the Docling library to handle the heavy lifting of PDF-to-structured-JSON conversion. The system supports three operational modes:

• Remote Mode: Connects to a Docling Serve API for lightweight processing.
• Local Mode: Performs all conversions locally on the user's machine.
• Hybrid Mode: Uses a remote service with an automatic fallback to local processing.

Who it’s for

Developers building RAG (Retrieval-Augmented Generation) applications or AI agents that need to ingest complex PDFs, generate structured documents, or integrate with vector databases like Milvus.

Highlights

• Agentic Document Processing: Turns PDF conversion and document generation into callable tools for LLMs.
• Flexible Deployment: Supports local, remote, or hybrid conversion modes.
• Multi-format Export: Converts PDFs to structured JSON and can export documents to Markdown.
• RAG Integration: Includes built-in support for LlamaIndex and LlamaStack for retrieval and extraction tasks.
• Performance Optimization: Features local document caching and memory management for large files.