Overview
What is Graphiti MCP Server?
Graphiti MCP Server is a knowledge graph server for AI agents, built with Neo4j and integrated with the Model Context Protocol (MCP). It enables dynamic graph management and semantic search, and is designed for developers using AI agents who need to store and query structured knowledge.
How to use Graphiti MCP Server?
Clone the repository, set up environment variables (including an OpenAI API key), then start the services with docker compose up. The server runs in Docker and exposes an SSE endpoint at http://localhost:8000/sse which can be configured in IDEs like Cursor by adding a JSON entry for MCP servers.
Key features of Graphiti MCP Server
- Dynamic knowledge graph management with Neo4j
- Seamless integration with OpenAI models
- MCP (Model Context Protocol) support
- Docker-ready deployment
- Custom entity extraction capabilities
- Advanced semantic search functionality
Use cases of Graphiti MCP Server
- Providing persistent knowledge memory for AI agent sessions
- Enabling semantic search over structured graph data
- Managing dynamic, evolving knowledge bases in agent workflows
FAQ from Graphiti MCP Server
What are the prerequisites to run Graphiti MCP Server?
You need Docker and Docker Compose, Python 3.10 or higher, and an OpenAI API key.
How do I start Graphiti MCP Server?
After cloning the repo and setting up the .env file, run docker compose up from the project directory.
Where is the knowledge graph data stored?
Data is stored in a Neo4j database that runs as part of the Docker Compose stack.
How can I integrate Graphiti MCP Server with Cursor IDE?
In Cursor’s MCP configuration, add a server entry with "url": "http://localhost:8000/sse". Optionally, apply user rules from graphiti_cursor_rules.md.
Which LLM models does Graphiti MCP Server work with?
It integrates with OpenAI models; the default model configured is gpt-4o.
