LLM Graph Builder MCP Server

LLM Graph Builder MCP

Build knowledge graphs from any URL using Claude Desktop and Neo4j.

What is this?

This Model Context Protocol (MCP) server enables Claude to automatically extract entities and relationships from unstructured text and build knowledge graphs in Neo4j. Simply give Claude a URL (Wikipedia article, PDF, web page, YouTube video) and ask it to build a knowledge graph - it handles the rest.

Perfect for: Research, Zotero integrations, academic papers, content analysis, and building structured knowledge from unstructured sources.

What's Included

This repository is a complete, ready-to-use package containing:

• llm_graph_builder_mcp/ - The MCP server code
• llm-graph-builder/ - Neo4j's LLM Graph Builder backend (June 24, 2025, commit 4d7bb5e8)

Both are included so you get a tested, working version out of the box. Just clone once and you're ready to go!

Why include the backend?

• Guaranteed compatibility - this MCP is tested with this exact backend version
• Zero configuration headaches - everything just works together
• If Neo4j updates their backend, you still have a working version

Features

• Multi-source support: Wikipedia, PDFs, web pages, YouTube videos
• Academic mode: Extract citations, authors, journals, and bibliographic data
• Custom schemas: Define allowed entity types and relationships
• Community detection: Find clusters and groups in your knowledge graph
• Zero modifications: Works with unmodified llm-graph-builder backend
• Local processing: Your data, your Neo4j instance, your control

Quick Start

Prerequisites

1. Neo4j database - Get a free instance at Neo4j AuraDB
   • Create an instance and note your connection URI, username, and password
2. OpenAI API key - Get one here
3. Python 3.10+ with uv - Install uv
4. Claude Desktop - Download here

Step 1: Clone This Repository

# Clone the entire project (includes both MCP and backend)
git clone https://github.com/henrardo/llm-graph-builder-mcp.git
cd llm-graph-builder-mcp

Your directory structure will be:

llm-graph-builder-mcp/           # The MCP server
llm-graph-builder/               # The backend (included)

Step 2: Set Up the Backend

# Navigate to backend
cd llm-graph-builder/backend

# Create environment file
cp example.env .env

Edit .env with your credentials:

# Neo4j Connection (from your AuraDB instance)
NEO4J_URI=neo4j+s://your-instance-id.databases.neo4j.io
NEO4J_USERNAME=neo4j
NEO4J_PASSWORD=your-auradb-password
NEO4J_DATABASE=neo4j

# OpenAI Configuration
LLM_MODEL_CONFIG_openai_gpt_4.1=gpt-4-turbo-2024-04-09,sk-your-openai-api-key

Install and start the backend:

# Create virtual environment
uv venv
source .venv/bin/activate  # Windows: .venv\Scripts\activate

# Install dependencies
uv pip install -r requirements.txt

# Start the backend server
uvicorn score:app --reload --port 8000

Keep this terminal running. The backend must be running for the MCP to work.

Step 3: Install the MCP

Open a new terminal (keep the backend running in the first one):

# Navigate back to the MCP directory
cd llm-graph-builder-mcp

# Install the MCP
uvx --from . llm-graph-builder-mcp

Step 4: Configure Claude Desktop

Edit your Claude Desktop config file:

• macOS: ~/Library/Application Support/Claude/claude_desktop_config.json
• Windows: %APPDATA%\Claude\claude_desktop_config.json

Add this configuration:

{
  "mcpServers": {
    "llm-graph-builder": {
      "command": "uvx",
      "args": [
        "--from",
        "/absolute/path/to/llm-graph-builder-mcp",
        "llm-graph-builder-mcp"
      ],
      "env": {
        "NEO4J_URI": "neo4j+s://your-instance-id.databases.neo4j.io",
        "NEO4J_USERNAME": "neo4j",
        "NEO4J_PASSWORD": "your-auradb-password",
        "NEO4J_DATABASE": "neo4j",
        "GRAPH_BUILDER_URL": "http://localhost:8000"
      }
    }
  }
}

Important:

• Replace /absolute/path/to/ with the full path to your llm-graph-builder-mcp directory
  • Run pwd in the llm-graph-builder-mcp directory to get this path
  • Example: /Users/yourname/projects/llm-graph-builder-mcp
• Use the same credentials as in your backend .env file

Step 5: Restart Claude Desktop

Completely quit and restart Claude Desktop for the changes to take effect.

Step 6: Test It

In Claude Desktop, try:

Build a knowledge graph from this Wikipedia article:
https://en.wikipedia.org/wiki/The_Hitchhiker%27s_Guide_to_the_Galaxy

Claude should now use the MCP to build a knowledge graph in your Neo4j database!

Usage Examples

Basic Usage

Build a knowledge graph from this Wikipedia