home / mcp / pdf rag mcp server

PDF RAG MCP Server

PDF RAG server for cursor.

Installation

Add the following to your MCP client configuration file.

Configuration

{
  "mcpServers": {
    "hyson666-pdf-rag-mcp-server": {
      "url": "http://localhost:7800/mcp"
    }
  }
}

You can connect your MCP-compatible tools and clients to a flexible knowledge base built around PDF processing, vector storage, and semantic search. This server exposes an MCP endpoint so AI tools can query processed PDFs, receive contextual results, and integrate with workflows like Cursor.

How to use

You use the MCP endpoint to access your PDF knowledge base from MCP-enabled tools. Two HTTP MCP endpoints are available: one for local testing and one for Cursor integration. You can point your MCP client to either endpoint to query processed PDFs, request semantically relevant chunks, and retrieve document context for AI workflows.

To connect via MCP, configure your MCP client with the appropriate server URL and start querying. The server provides real-time processing status updates and semantic search results, so you receive feedback as documents are uploaded, processed, and indexed.

Two explicit MCP endpoints are shown for integration. The first endpoint is intended for general MCP access, and the second is commonly used when configuring Cursor to talk to the server.

{
  "mcpServers": {
    "pdf_rag": {
      "url": "http://localhost:7800/mcp"
    }
  }
}

How to install

Prerequisites you need before starting: Python 3.8 or later, uvx for Python dependency management, Git, and optionally Cursor for MCP integration.

Step-by-step setup and startup flow you can follow to run the server locally and access the web interface.

# 1. Clone the project
git clone https://github.com/yourusername/PdfRagMcpServer.git
cd PdfRagMcpServer

# 2. Install uv if you don't have it
curl -sS https://astral.sh/uv/install.sh | bash

# 3. Create and activate a virtual environment
uv venv
source .venv/bin/activate

# 4. Install backend dependencies
uv pip install -r backend/requirements.txt

# 5. Start the application
uv run run.py

# 6. Open the web interface
# http://localhost:8000

MCP endpoints and integration notes

You can integrate with Cursor by adding the server as a Custom MCP Server. Use the following configuration to connect Cursor to the MCP endpoint.

{
  "mcpServers": {
    "pdf_rag": {
      "url": "http://localhost:7800/mcp"
    }
  }
}

Cursor integration example for MCP URL

If you are configuring Cursor to use the server over MCP, set the URL to the server's MCP endpoint as shown.

{
  "mcpServers": {
    "pdf_rag": {
      "url": "http://localhost:8000/mcp/v1"
    }
  }
}

Available tools

pdf_processing

Uploads PDFs, extracts text, chunks content, and vectorizes embeddings for semantic search.

real_time_status

Provides WebSocket-based real-time updates on document processing progress.

semantic_search

Performs vector-based semantic search across processed PDFs to retrieve relevant passages.

mcp_integration

Exposes an MCP server endpoint to enable integration with MCP clients like Cursor.

frontend_interface

Delivers a modern React/Chakra UI frontend for document management and querying.