madeinoz-knowledge-system

This MCP server provides a persistent personal knowledge management system built on a Graphiti knowledge graph with optional backends (Neo4j or FalkorDB). It automatically extracts entities, discovers relationships, and enables semantic search across conversations and documents. With this MCP pack, you can capture knowledge by natural language prompts, perform semantic queries to locate related concepts, and run investigative searches to explore interconnected memories within configurable hops. Temporal tracking lets you see how knowledge evolves over time, and memory decay scoring helps prioritize important information. The included server tooling supports management tasks such as start, stop, status, and logs, making it easy to operate within your AI infrastructure.

Prerequisites:

• Docker and Docker Compose installed on your host
• Basic familiarity with running containers

Installation steps:

1. Pull and run the MCP server container (example):

# Ensure Docker is running
docker version

# Run the knowledge system container (configured for Neo4j as backend by default)
docker run -d --name madeinoz-knowledge-system \
  -e GRAPH_BACKEND=neo4j \
  -e NEO4J_URL=bolt://localhost:7687 \
  -e NEO4J_USER.neo4j \
  -e NEO4J_PASSWORD=your-neo4j-password \
  ghcr.io/madeinoz67/madeinoz-knowledge-system:latest

2. If you prefer Docker Compose, create a docker-compose.yml with equivalent environment variables and run:

version: '3.8'
services:
  knowledge-system:
    image: ghcr.io/madeinoz67/madeinoz-knowledge-system:latest
    environment:
      GRAPH_BACKEND: neo4j
      NEO4J_URL: bolt://neo4j:7687
      NEO4J_USER: neo4j
      NEO4J_PASSWORD: your-neo4j-password
      LOG_LEVEL: info
    ports:
      - "7474:7474" # Neo4j UI
      - "3000:3000" # FalkorDB UI

3. Verify installation:

docker ps
curl -i http://localhost:7474/ 
curl -i http://localhost:3000/ 

4. Configure environment specifics per your deployment (database credentials, endpoints, and backends) before production use.

Tips and common considerations:

• Backends: Neo4j is the recommended default for rich querying, while FalkorDB offers simpler setup and lower resource usage. Ensure the chosen backend is reachable from the MCP container.
• Environment variables: Securely manage credentials (prefer secrets management in production). Example vars include NEO4J_PASSWORD and GRAPH_BACKEND.
• Monitoring: Prometheus and Grafana dashboards are supported via the included architecture references; ensure ports are exposed if you want to visualize metrics.
• Networking: When running behind a NAT or in Kubernetes, adjust URLs (NEO4J_URL, FalkorDB endpoints) accordingly and consider using TLS.
• Data safety: Regular backups of the graph database are recommended; coordinate backup scripts with your backend choice.
• Troubleshooting: If the container fails to start, check container logs (docker logs <container>) and verify that the backend service (Neo4j/FalkorDB) is up and accessible.