Knowledge Graph Memory Server

Example Use Cases

• 1

  Persist user identity, preferences and contextual facts across conversations to provide personalized responses.

• 2

  Record, search, and retrieve structured knowledge (entities, relations, observations) for context-aware decision making.

• 3

  Capture recurring errors and verified solutions (lessons), recommend fixes, and track solution success rates over time.

Description

This MCP server implements a persistent local knowledge graph for storing entities, directed relations, and atomic observations, plus a lesson system for capturing errors and their verified solutions. It supports CRUD operations on entities, relations, observations and specialized lesson management tools (create, find similar errors, update success, and get recommendations). Files are managed efficiently (memory.json and lesson.json with automatic splitting) and the server is designed for easy integration with Cursor MCP clients and Claude Desktop. The project is open-source (MIT) and intended to help agents personalize interactions and improve over time by learning from past errors.

Capabilities(9 total)