MCP Neo4j

What is MCP Neo4j about?

MCP Neo4j provides a set of servers that translate natural language commands from LLM‑powered clients into Cypher queries, manage Neo4j Aura cloud instances, store personal knowledge graphs, and support interactive data‑modeling. All servers adhere to the Model Context Protocol, offering consistent communication across clients such as Claude Desktop, VS Code, Cursor, and Windsurf.

How to use MCP Neo4j?

1. Select a server – Choose one of the four provided servers depending on the required capability (Cypher execution, knowledge‑graph memory, Aura management, or data‑modeling).
2. Run the server – Start the server in the desired transport mode (STDIO, SSE, or HTTP). For HTTP, add --transport http plus optional host, port, and path flags or set the corresponding environment variables.
3. Connect a client – Configure your MCP‑compatible client to point at the server endpoint. The client will send natural‑language prompts, which the server converts into Neo4j operations.
4. Interact – Ask questions, request visualizations, create or scale Aura instances, or store personal facts; the server returns results in real‑time.

Key features of MCP Neo4j

• Natural‑language to Cypher – mcp-neo4j-cypher generates and runs read/write queries based on conversational input.
• Graph memory store – mcp-neo4j-memory persists entities and relationships, accessible across sessions.
• Aura cloud management – mcp-neo4j-cloud-aura-api creates, scales, and deletes Aura instances, and toggles features such as Graph Data Science.
• Interactive data modeling – mcp-neo4j-data-modeling validates, visualizes, and imports/exports models (e.g., from Arrows.app).
• Multiple transport modes – STDIO for local tooling, SSE for event‑driven web apps, and streamable HTTP for microservice deployments.
• Environment‑driven configuration – Host, port, and path can be supplied via NEO4J_TRANSPORT, NEO4J_MCP_SERVER_HOST, NEO4J_MCP_SERVER_PORT, and NEO4J_MCP_SERVER_PATH.

Use cases of MCP Neo4j

• Exploratory data analysis – Ask “What are the top‑selling products?” and receive a chart generated from Cypher results.
• Knowledge‑graph augmentation – Store facts like “Worked on Neo4j MCP Servers with Andreas and Oskar” for later retrieval.
• Automated cloud provisioning – Create a new Aura instance with a single natural‑language command.
• Collaborative modeling – Teams can co‑design graph schemas and instantly visualize them.
• LLM‑powered assistants – Embed Neo4j capabilities into AI assistants without writing code.

FAQ from the MCP Neo4j project

Q: Which transport mode should I choose? A: Use STDIO for local tools or Claude Desktop. Choose HTTP for web‑based deployments or when integrating with other microservices. SSE is useful for server‑sent event streams.

Q: Do I need a Neo4j instance for every server? A: mcp-neo4j-cypher and mcp-neo4j-memory require a configured Neo4j database (local or remote). The Aura API server interacts directly with Neo4j’s cloud platform.

Q: How are authentication credentials provided? A: Pass Neo4j connection URIs, usernames, and passwords via standard environment variables (e.g., NEO4J_URI, NEO4J_USER, NEO4J_PASSWORD). Aura API keys can be set with NEO4J_AURA_TOKEN.

Q: Can I run multiple servers simultaneously? A: Yes. Each server can listen on a different port or path; configure them individually using the --host, --port, and --path flags.

Q: Is the project open source? A: The repository is licensed under the MIT License, and contributions are welcomed via pull requests.

Neo4j MCP Clients & Servers

Model Context Protocol (MCP) is a standardized protocol for managing context between large language models (LLMs) and external systems.

This lets you use Claude Desktop, or any other MCP Client (VS Code, Cursor, Windsurf), to use natural language to accomplish things with Neo4j and your Aura account, e.g.:

• What is in this graph?
• Render a chart from the top products sold by frequency, total and average volume
• List my instances
• Create a new instance named mcp-test for Aura Professional with 4GB and Graph Data Science enabled
• Store the fact that I worked on the Neo4j MCP Servers today with Andreas and Oskar

Servers

mcp-neo4j-cypher - natural language to Cypher queries

Details in Readme

Get database schema for a configured database and execute generated read and write Cypher queries on that database.

mcp-neo4j-memory - knowledge graph memory stored in Neo4j

Details in Readme

Store and retrieve entities and relationships from your personal knowledge graph in a local or remote Neo4j instance. Access that information over different sessions, conversations, clients.

mcp-neo4j-cloud-aura-api - Neo4j Aura cloud service management API

Details in Readme

Manage your Neo4j Aura instances directly from the comfort of your AI assistant chat.

Create and destroy instances, find instances by name, scale them up and down and enable features.

mcp-neo4j-data-modeling - interactive graph data modeling and visualization

Details in Readme

Create, validate, and visualize Neo4j graph data models. Allows for model import/export from Arrows.app.

Transport Modes

All servers support multiple transport modes:

• STDIO (default): Standard input/output for local tools and Claude Desktop integration
• SSE: Server-Sent Events for web-based deployments
• HTTP: Streamable HTTP for modern web deployments and microservices

HTTP Transport Configuration

To run a server in HTTP mode, use the --transport http flag:

# Basic HTTP mode
mcp-neo4j-cypher --transport http

# Custom HTTP configuration
mcp-neo4j-cypher --transport http --host 127.0.0.1 --port 8080 --path /api/mcp/

Environment variables are also supported:

export NEO4J_TRANSPORT=http
export NEO4J_MCP_SERVER_HOST=127.0.0.1
export NEO4J_MCP_SERVER_PORT=8080
export NEO4J_MCP_SERVER_PATH=/api/mcp/
mcp-neo4j-cypher

Contributing

Contributions are welcome! Please feel free to submit a Pull Request.

Blog Posts

• Everything a Developer Needs to Know About the Model Context Protocol (MCP)
• Claude Converses With Neo4j Via MCP - Graph Database & Analytics
• Building Knowledge Graphs With Claude and Neo4j: A No-Code MCP Approach - Graph Database & Analytics

License

MIT License