Memory MCP Server logo

Memory MCP Server

by okooo5km

The Memory MCP Server is a Model Context Protocol server that provides knowledge graph management capabilities. It enables LLMs to maintain memory across conversations by creating, reading, updating, and deleting entities and relations in a persistent knowledge graph.

View on GitHub

Last updated: N/A

What is Memory MCP Server?

The Memory MCP Server is a Swift implementation of a Model Context Protocol (MCP) server designed to provide knowledge graph management capabilities for Large Language Models (LLMs). It allows LLMs to create, read, update, and delete entities and relations in a persistent knowledge graph, enabling them to maintain memory and context across conversations.

How to use Memory MCP Server?

The server can be installed either using a one-line curl command or by building from source. Once installed, it can be configured with tools like Claude.app, Cursor, and Chatwise by adding the server's command and environment variables (specifically MEMORY_FILE_PATH) to their respective settings. The server exposes several tools for managing the knowledge graph, including creating entities and relations, adding observations, deleting entities and relations, reading the graph, and searching nodes. These tools are accessed via JSON payloads.

Key features of Memory MCP Server

  • Knowledge Graph Storage

  • Entity Management

  • Relation Tracking

  • Observation System

  • Powerful Search

  • Persistent Storage

Use cases of Memory MCP Server

  • Long-term Memory for AI Assistants

  • Knowledge Management

  • Relationship Tracking

  • Context Persistence

  • Journal and Daily Logs

FAQ from Memory MCP Server

What is the default storage location for the knowledge graph?

If not specified by the MEMORY_FILE_PATH environment variable, the knowledge graph is stored in memory.json in the current working directory.

How do I configure the server for use with Claude.app?

Add the server configuration to your Claude settings, including the command to run the server and the MEMORY_FILE_PATH environment variable.

What is the purpose of the search_nodes tool?

The search_nodes tool allows you to search for nodes in the knowledge graph based on a query that matches against entity names, types, and observations.

What is the structure of the knowledge graph?

The knowledge graph consists of entities (nodes) with a name, type, and list of observations, and relations (edges) between entities with a relation type in active voice. The graph is persisted as a line-delimited JSON file.

What are the development requirements?

You need Swift 6.0 or later, macOS 14.0 or later, and MCP Swift SDK 0.2.0 or later.