Neo4j MCP SSE Bridge

Production-ready Server-Sent Events (SSE) bridge for Model Context Protocol (MCP) with Neo4j knowledge graph functionality, designed for deployment at memory.aynshteyn.dev/mcp.

Features

• Knowledge Graph Management: Create, read, update, and delete entities and relationships
• Memory Persistence: Store conversation context and insights across sessions
• Cypher Query Execution: Direct Neo4j database access for advanced operations
• SSE Connectivity: Real-time communication with MCP clients over HTTPS
• Production Security: SSL/TLS encryption and secure authentication
• Domain-Ready: Pre-configured for memory.aynshteyn.dev deployment

Quick Start

1. Install dependencies:

pip install -r requirements.txt

2. Configure environment variables in .env file

3. Run the server:

python main.py

Production Deployment

The server is designed for deployment at https://memory.aynshteyn.dev/mcp

• Development: Runs on HTTP port 8000 without SSL
• Production: Runs on HTTPS port 443 with SSL certificates
• Transport: SSE (Server-Sent Events) for real-time communication
• Path: Serves MCP functionality on /mcp endpoint

SSL Certificate Requirements

For production deployment with SSL, ensure the SSL certificate files exist at the paths specified in your .env file:

• SSL_CERTFILE must point to a valid SSL certificate file
• SSL_KEYFILE must point to the corresponding private key file

The server automatically detects SSL certificate availability and falls back to HTTP for development if certificates are not found.

Core Functionality

Entity Management

• Create and organize knowledge entities with types and observations
• Establish relationships between entities using active voice descriptions
• Search and retrieve entities by name, type, or content

Knowledge Graph Operations

• Full graph reading and visualization
• Targeted entity searches and filtering
• Bulk operations for efficient data management

Database Integration

• Direct Cypher query execution for advanced operations
• Secure Neo4j connection with authentication
• Support for both local and cloud Neo4j instances

STDIO Servers

The servers/ directory contains standalone MCP servers that operate over STDIO:

• mcp-neo4j-memory: Knowledge graph memory storage
• mcp-neo4j-cypher: Natural language to Cypher translation
• mcp-neo4j-cloud-aura-api: Neo4j Aura cloud management

These servers can be used independently with any MCP client that supports STDIO communication.

License

MIT License