MCP PDF Server provides a Python-based backend for managing and querying PDF datasheets via MCP (Model Context Protocol). It consists of two main components: a manager server (FastAPI web UI and RESTful API for uploading, listing, and downloading PDFs) and an MCP server that extracts text from PDFs and exposes MCP-compatible tools to external systems such as Cursor. The available MCP tools include reading text from local PDFs (read_local_pdf), reading text from PDF URLs (read_url_pdf), listing all PDFs under the datasheets directory (server_pdf_list), and searching PDFs by filename to retrieve and extract their text (server_pdf_search). You can access the web UI and REST API to upload PDFs to /app/datasheets, manage your files, and use the MCP endpoints to integrate with external tooling. The server is designed to run either locally (Python) or in Docker, so you can deploy in a development or production environment and connect to external systems via MCP.

To start, configure the MCP server through the mcp-pdf-server entry; the MCP interface will expose tools for external systems to invoke text extraction and search against the stored PDFs. When interacting with MCP tooling, you can perform a search by filename to locate a relevant PDF and then extract its text for downstream AI tasks or data ingestion workflows.

Prerequisites:

• Python 3.8+ (or a compatible Python environment)
• Pip (preferred via Python installation)
• Optional: Docker (if you prefer containerized deployment)

Local installation (Python):

1. Clone the repository: git clone https://github.com/your-organization/mcp-pdf-server.git cd mcp-pdf-server

2. Create and activate a virtual environment (recommended): python -m venv venv source venv/bin/activate # Linux/macOS venv\Scripts\activate.bat # Windows

3. Install dependencies: pip install -r requirements.txt

4. Run the MCP PDF Server (MCP mode will be available via MCP tooling): uvicorn manager_server.main:app --host 0.0.0.0 --port 5080

   or start the mcp server module if configured differently by your environment

Docker (alternative):

1. Build the image: docker build -t mcp-pdf-server:1.0.0 .

2. Run container (mount your PDFs to /app/datasheets inside the container): docker run -d
   -v /path/to/your/datasheets:/app/datasheets
   -p 5050:5050
   -p 5080:5080
   --name mcp-pdf-server
   mcp-pdf-server:1.0.0

3. Optional: use docker-compose (adjust the /path/to/your/datasheets to your actual folder): docker-compose up -d --build

Notes and tips:

• The datasheets must reside under /app/datasheets inside the running environment. When using Docker, mount your host PDF folder to that path.
• The server exposes ports 5050 and 5080 for the web API and MCP interactions; ensure these ports are available on your host.
• If you modify dependencies, update requirements.txt and reinstall for your environment.
• For MCP integration, use the read_local_pdf and read_url_pdf tools to fetch text, then use server_pdf_list and server_pdf_search to navigate and retrieve content efficiently.
• In production, consider setting up authentication for the web API and enabling HTTPS where appropriate.