This project provides a local knowledge graph server for AI models, enabling persistent memory across conversations. It's designed for users who want to manage AI memory locally, offering a customizable data path and compatibility with any AI model supporting the Model Context Protocol (MCP) or function calling.

How It Works

The server structures persistent memory as a knowledge graph composed of entities, relations, and observations. Entities represent distinct concepts (people, places, events), linked by directed relations (e.g., "works_at"). Observations are atomic facts attached to entities. This graph-based approach allows for rich, interconnected memory storage and retrieval, facilitating more contextually aware AI interactions.

Quick Start & Requirements

• Install: npx -y mcp-knowledge-graph
• Prerequisites: Node.js (for npx).
• Configuration: Requires integration with AI platforms via MCP or function calling. Example configuration provided for Claude Desktop.
• Docs: README

Highlighted Details

• Supports local, persistent memory storage via a customizable file path (--memory-path).
• Provides a RESTful API for CRUD operations on entities, relations, and observations.
• Includes search functionality to query the knowledge graph by entity names, types, or observation content.
• Designed for integration with various AI models (Claude, GPT, Llama) via MCP or function calling.

Maintenance & Community

The project is a fork of an original Memory Server, with a focus on local development. Community and maintenance details are not explicitly provided in the README.

Licensing & Compatibility

Licensed under the MIT License, permitting commercial use and integration with closed-source applications.

Limitations & Caveats

The README indicates this is a fork focused on local development, but does not detail specific differences or potential limitations compared to the original project. The setup instructions primarily use npx, implying a Node.js environment is required.