Neo4j Knowledge Graph Memory

Neo4j Knowledge Graph Memory is an MCP server that connects large language model interfaces to a Neo4j graph database to store, update, and retrieve contextual information across conversations. Built as an enhanced implementation of the official Model Context Protocol knowledge graph memory specification, it replaces flat JSON files or local tabular databases with a native graph engine. Developers, knowledge engineers, and AI practitioners use this server to build complex personal knowledge graphs, agentic memory layers, and retrieval-augmented generation pipelines. It enables AI clients to perform complete create, read, update, and delete operations on entities, relationships, and observations. The server pairs Neo4j full-text indexes with Fuse.js for fuzzy and exact matching, allowing assistants to recall interconnected user preferences, professional contacts, and organizational facts. Because it runs on Neo4j, developers can scale knowledge representations to enterprise workloads, run advanced Cypher graph queries, and visually inspect memory clusters using native graph database tools.

Category: AI Memory & Context

Tags: graph database, knowledge-graph, memory, neo4j, rag

Visit Neo4j Knowledge Graph Memory

How to install and configure Neo4j Knowledge Graph Memory

1. Ensure Node.js (version 22.0.0 or higher) and a running Neo4j instance are installed, or run the project using Docker Compose: bash git clone https://github.com/JovanHsu/mcp-neo4j-memory-server.git cd mcp-neo4j-memory-server docker-compose up -d 2. Open your claude_desktop_config.json file. 3. Add the server under the mcpServers object with your database connection credentials: json { "mcpServers": { "graph-memory": { "command": "npx", "args": [ "-y", "@izumisy/mcp-neo4j-memory-server" ], "env": { "NEO4J_URI": "neo4j://localhost:7687", "NEO4J_USER": "neo4j", "NEO4J_PASSWORD": "password", "NEO4J_DATABASE": "memory" } } } } 4. Restart Claude Desktop to enable the knowledge graph memory integration.

What you can do with Neo4j Knowledge Graph Memory

• Storing conversational facts, user traits, and behavioral preferences across ongoing AI dialogue sessions as interconnected graph nodes. - Mapping professional and personal relationships between individuals, groups, and organizations up to multiple degrees of separation. - Performing fuzzy and exact searches over stored contextual observations using integrated Neo4j indexes and Fuse.js algorithms. - Managing full CRUD cycles for knowledge graph entities and directed edges via standard Model Context Protocol client calls. - Visualizing conversational memory structures and relationship webs directly inside the native Neo4j Browser or Bloom interfaces.

Key facts

• Open Source
• https://github.com/JovanHsu/mcp-neo4j-memory-server
• AI Memory & Context, Databases & Data Stores
• graph database, knowledge-graph, memory, neo4j, rag

Part of MCP Servers

How do I install Neo4j Knowledge Graph Memory?

You can install it globally using npm install -g @jovanhsu/mcp-neo4j-memory-server or run it directly via npx. You can also clone the repository and deploy the database and memory server together using docker-compose up -d.

What can Neo4j Knowledge Graph Memory do?

It provides persistent memory management for AI models by storing entities, observations, and relationships inside a Neo4j graph database. It supports fuzzy search, exact matching, relationship traversal, and complete CRUD operations over graph nodes.

Which MCP clients work with Neo4j Knowledge Graph Memory?

It works with Claude Desktop, the MCP Inspector, and custom TypeScript or Node.js applications that communicate using the official Model Context Protocol SDK via standard input and output transport.

Is Neo4j Knowledge Graph Memory open source?

Yes, Neo4j Knowledge Graph Memory is open source and distributed under the MIT license on GitHub.