AI-powered graph database access for LLM assistants
From $65/mo (free plan available)FreemiumDeveloper Tools
Visit Website FreemiumVisit Website
Tracked since2026
The Bottom Line
- Price
- From $65/mo (free plan available). Compare plans
Key facts
- Official MCP server that connects LLM assistants to any Neo4j graph database deployment
- Supports graph schema retrieval, Cypher queries (read/write), and graph algorithm execution via GDS
- Works with multiple MCP hosts (Claude, VS Code, Cursor, Copilot) and transport protocols (stdio, HTTP)
Pros
- Official Neo4j-maintained server ensures reliability and compatibility
- Enables natural language querying of graph databases without writing Cypher manually
- Integrates with popular AI IDEs and agent frameworks for streamlined development workflows
Cons
- Still early-stage with open security and permission control issues in the MCP protocol
- Requires local setup and configuration of environment variables for each MCP host
What is Neo4j MCP?
Neo4j MCP is the official Model Context Protocol server for Neo4j, enabling AI assistants and LLM-powered tools to interact directly with Neo4j graph databases. It acts as a bridge between MCP hosts (like Claude Desktop, VS Code, Cursor, and Copilot) and any Neo4j deployment, whether AuraDB, self-managed, Docker, or sandbox. The server exposes capabilities such as retrieving graph schemas, executing read and write Cypher queries, and running graph algorithms via the Graph Data Science plugin.
Designed for developers and data scientists, Neo4j MCP simplifies building intelligent applications that leverage connected data. It supports both stdio and HTTP transport (including custom SSL certificates) and integrates seamlessly with popular AI-powered IDEs and agent frameworks. The server is open source, downloadable as a binary, and maintained by Neo4j Labs, making it a reliable choice for embedding graph database interactions into AI workflows.
Preview
Key Features
- Graph schema retrieval for understanding database structure
- Execute read and write Cypher queries directly from AI assistants
- Run graph algorithms (centrality, pathfinding, etc.) via the GDS plugin
- Supports stdio and HTTP transport with custom SSL certificates
- Compatible with all Neo4j deployment types: AuraDB, self-managed, Docker, Desktop, Sandbox
- Open source binary available on GitHub with installation guides for common MCP hosts
Pricing Plans
Free TrialPricing checked Oct 1, 2026
| Plan | Price | Details |
|---|---|---|
| AuraDB Free | Free |
+13 moreShow less
|
| AuraDB Professional | $65 / GB/month* (minimum 1GB cluster) |
+33 moreShow less
|
| AuraDB Business Critical | $146 / GB/month* (minimum 2GB cluster) |
+39 moreShow less
|
| AuraDB Virtual Dedicated Cloud | Contact us |
+3 moreShow less
|
Is Neo4j MCP worth the price?
Fair value
Neo4j's AuraDB pricing is fair for a managed graph database with AI integration capabilities, though the Professional tier at $65/GB/month can become expensive for large datasets.
The Free tier is generous for learning and prototyping, but production costs scale quickly with memory needs. Best for teams that need graph-powered context for LLMs and can justify the premium over simpler vector stores.
Hidden Costs & Gotchas
Minimum 1GB cluster at Professional tier
Business Critical requires minimum 2GB cluster
Reviews
$99Free with your review
Write a reviewReview Neo4j MCP, get a free AI guide
Share your experience and we will send you Improve Your Thinking Patterns Using ChatGPT, free.
Explore More
Neo4j MCP FAQ
Which MCP hosts are compatible with the Neo4j MCP server?
The Neo4j MCP server works with any MCP host that supports stdio or HTTP transport, including Claude Desktop, VS Code, Cursor, Windsurf, Cline, and Microsoft Copilot Studio. OpenAI ChatGPT Desktop support is coming soon.
Can I use the Neo4j MCP server with Neo4j AuraDB?
Yes, the official MCP server works with all Neo4j deployments, including Neo4j AuraDB, self-managed instances, Docker containers, Neo4j Desktop, and Neo4j Sandbox.
Does the Neo4j MCP server support graph algorithms?
Yes, it supports executing graph algorithms from the Neo4j Graph Data Science (GDS) plugin, including projections, enabling tasks like centrality analysis, community detection, and pathfinding directly from AI assistants.
What transport protocols does the Neo4j MCP server offer?
The server supports both stdio (standard input/output) and HTTP transport. HTTP transport includes support for custom SSL certificates and recently added OAuth 2.1 authentication.
How do I install the Neo4j MCP server?
The server is distributed as a downloadable binary from the GitHub repository. Installation instructions include setting environment variables for common MCP hosts like Claude Desktop and VS Code, and there is a guide for deploying on GCP Cloud Run.
Is the Neo4j MCP server open source?
Yes, the official Neo4j MCP server is open source and hosted on GitHub under the Neo4j Labs organization. The repository includes the source code, issue tracker, and documentation.
Can the Neo4j MCP server be used with agent frameworks like LangChain or CrewAI?
Yes, agent frameworks such as LangChain, CrewAI, Pydantic.AI, and the Agent Development Kit (ADK) support MCP as tools, so they can integrate the Neo4j MCP server to give AI agents graph database access.
What are the limitations of the Neo4j MCP server regarding security?
The MCP protocol has open issues around security, discovery, trusted deployment, and permission control. However, Anthropic and the community are actively addressing these, and large providers joining the ecosystem are expected to resolve enterprise requirements soon.
Source: neo4j.com