MCP Rawr Graphiti

The Graphiti MCP Server is a multi - project knowledge graph server extension based on the official getzep/graphiti. It supports multi - server single - database configuration, provides a developer - friendly CLI tool, and can build temporary knowledge graphs for each project for AI agents to query through the model context protocol.

Knowledge management and memory Developer tools #Knowledge Graph #Multi - project #AI Agent #Neo4j .Python

rating : 2 points

downloads : 13.0K

update time : 2025-04-29

Open Site

What is the Graphiti MCP Server?

The Graphiti MCP Server is a powerful tool for transforming unstructured text into time - series knowledge graphs. It extracts entities and relationships through language models and records each change as a timestamp event, thus supporting versioned queries.

How to use the Graphiti MCP Server?

You can get the Graphiti MCP Server up and running in just a few simple steps. First, install the CLI tool, then generate the Docker Compose file, next start the service, and finally initialize your project.

Application Scenarios

Suitable for application scenarios that require managing and querying knowledge graphs across multiple projects, such as enterprise knowledge management and AI assistant development.

Main Features

Multi - project Support

Allows multiple projects to share the same Neo4j database while maintaining complete isolation.

Time - series Graph

Records the timestamp of each change and supports versioned queries.

Automatic Discovery and Configuration

Automatically generates the MCP configuration file required by the IDE, simplifying the integration process.

Advantages

Supports multi - project isolation to avoid conflicts.

Provides zero - downtime maintenance capabilities.

Easy to integrate into existing IDEs and AI frameworks.

Limitations

May seem overly complex for small projects.

Requires a certain technical background for setup and maintenance.

How to Use

Install the CLI Tool

Install the CLI tool for the Graphiti MCP Server via pipx.

Generate the Docker Compose File

Run the graphiti compose command to generate the Docker Compose configuration file.

Start the Service

Use the graphiti up -d command to start the service.

Initialize a New Project

Run the graphiti init command in the root directory of the project to initialize a new project.

Usage Examples

Create a Knowledge Graph

Create a knowledge graph using the Graphiti MCP Server.

Start the Service

Start the service of the Graphiti MCP Server.

Frequently Asked Questions

How to ensure isolation between multiple projects?

Can I run only one server?

How to safely clear the entire knowledge graph?

Related Resources

Official Documentation

The official documentation of the Graphiti MCP Server.

GitHub Repository

The open - source code repository of the Graphiti MCP Server.

Community Forum

The user community forum of the Graphiti MCP Server.

🚀 Graph Representation Graph

Graph Representation Graph is a project that provides a set of tools and configurations for managing and running Graphiti MCP services.

🚀 Quick Start

1. Clone the Repository and Configure the Environment

git clone https://github.com/rawr-ai/mcp-graphiti.git
cd mcp-graphiti
cp .env.example .env   # Copy the example environment file

⚠️ Important Note

If NEO4J_PASSWORD is set to 'password', the service will refuse to start unless GRAPHITI_ENV=dev is explicitly set.

Use strong passwords except in the development environment.

2. Install the CLI Tool

pipx install . --include-deps   # Recommended for users
# Or for contributors:
python -m venv .venv && source .venv/bin/activate && pip sync uv.lock && pip install -e .

3. Start the Service

graphiti compose   # Generate the docker-compose file
graphiti up -d     # Start the service

💻 Usage Examples

Create a Project

cd ~/code
graphiti init acme-support-bot   # Run at the root of the new project
cd acme-support-bot
# Add entity YAML files to the ai/graph/entities/ directory

📚 Documentation

Project Structure

mcp-graphiti/
├── .env.example
├── docker-compose.yml
├── mcp-projects.yaml
├── README.md
└── graphiti/
    ├── main.py
    └── ...

Project Configuration (`mcp-projects.yaml`)

This file is used to define and manage multiple Graphiti MCP service instances.

💡 Usage Tip

List all projects, specifying the root directory path and other configurations.

By default, mcp-projects.yaml is added to the .gitignore file to avoid accidental commits to version control.

For private project management, remove mcp-projects.yaml from the local .gitignore file.

Run graphiti compose to generate the docker-compose.yml configuration file.

Single Service vs. Multiple Services FAQ

Question	Answer
Can it be merged into one service?	Yes, remove projects from `mcp-projects.yaml` or set `MCP_SINGLE_SERVER=true` and re-run `compose`.
Is isolation only achieved through ports?	No, each query contains a `group_id`; additional containers provide crash and dependency isolation.
Can a gateway be added?	Of course, any API gateway or reverse proxy can inject the `group_id` and route to the root service.

Danger Zone

Setting NEO4J_DESTROY_ENTIRE_GRAPH=true will erase all project data. Use with caution.

Roadmap and Contribution

RAWR CLI Integration: Expose the entire RAWR stack through a top-level command.
graphiti prune: Recover space for orphaned group_id graphs with a single command.

PRs and issues are welcome!