MCP PDF Server

MCP PDF Server is a Model Context Protocol (MCP) based server that can efficiently manage PDF files.

This project was created to make it easier for me, an embedded developer, to read and summarize PDF datasheet documents directly in AI coding tools such as Cursor, or to conduct Q&A, etc., to make development work more convenient. In other words, the main purpose is to support AI to quickly understand the contents of PDF datasheets and provide the necessary information right away.

This project consists of two main components:

• manager_server : A web page based on FastAPI that allows users to upload or download PDF files and view and manage file lists through a web UI. It also provides a RESTful API for linking with external systems.

• mcp_server : Provides file name search and text extraction functions based on PDF files managed by manager_server. The extracted text can be linked with external systems (e.g. Cursor, etc.) via the MCP protocol.

Key Features:

• Extract PDF text (supports local files and URLs)

• PDF search based on file name

• View and manage PDF lists

• Supports PDF file web upload/download

• Providing RESTful APIs and web services

• Connectivity to external systems (Curator, Cursor, etc.) via MCP protocol

It can be easily integrated with external systems through RESTful API and web UI, and can be easily deployed and operated in both Docker and local environments. It is suitable for automated management and search of various PDF documents such as data sheets, papers, and contracts.

Key Features

• Extract text from local PDF files and PDFs accessible via URL

• Provides a list of PDF files under /app/datasheets

• Provides PDF search function by file name

• Reliable text extraction and exception handling based on PyPDF2

• Providing standardized MCP tools based on FastMCP

Related MCP server: pdf-mcp

Running with Docker

1. Build image

   docker build -t mcp-pdf-server:1.0.0 .

2. Running a container

   docker run -d \
     -v /호스트/경로/데이터:/app/datasheets \
     -p 5050:5050 \
     -p 5080:5080 \
     --name mcp-pdf-server \
     mcp-pdf-server:1.0.0

   • If you put a PDF file in /호스트/경로/데이터 you can access it from /app/datasheets inside the container.

   • Ports 5050 and 5080 are used.

3. When using docker-compose

   # docker-compose.yml의 /path/to/your/datasheets를 실제 PDF 폴더 경로로 수정하세요.
   docker-compose up -d --build

Running it directly locally (Python)

1. Installing dependencies

   pip install -r requirements.txt

2. Running the server

   python mcp_server/mcp_pdf_server.py
   # 또는
   uvicorn manager_server.main:app --host 0.0.0.0 --port 5080

MCP Tool (API) Description

• read_local_pdf
  Extract text by entering a local PDF file path.

• read_url_pdf
  Enter a URL of a PDF file and extract the text from it.

• server_pdf_list
  Returns a list of all PDF files under /app/datasheets .

• server_pdf_search
  Enter a file name, search for a PDF file on the server, and extract the text from that PDF file.

Route guidance

• PDF data must be located in the /app/datasheets path (inside the Docker container).

• When using Docker, mount the host's PDF folder to /app/datasheets .

• The source code is located in /app/mcp_server (inside the container).

License

Apache License 2.0
Author: Dev91