home / mcp / mcp knowledge graph mcp server

MCP Knowledge Graph MCP Server

Provides a local knowledge graph memory store for AI models with master and named databases accessible via MCP tools.

Installation

Add the following to your MCP client configuration file.

Configuration

{
  "mcpServers": {
    "shaneholloman-mcp-knowledge-graph": {
      "command": "npx",
      "args": [
        "-y",
        "mcp-knowledge-graph",
        "--memory-path",
        "/Users/yourusername/.aim/"
      ]
    }
  }
}

You can run a local AI memory graph that persists across conversations, enabling your AI to store entities, relations, and observations for quick retrieval. This MCP server keeps memory centralized and safe, while letting you choose between project-local storage or a global memory directory.

How to use

Set up a memory server to store and access your AI knowledge graph. Use a local project memory when you are working within a specific project, or configure a global memory location for all projects. Your AI can create entities, link them with relations, and add observations to build a persistent graph you can query later.

How to install

Prerequisites you need before starting: - Node.js 18+ installed on your system - An MCP-compatible AI platform ready to connect to memory data Follow these steps to install and run the memory graph server using the MCP setup shown here.

1) Ensure Node.js is installed. Check the version: ``node -v`` If it's not installed, download and install it from the official Node.js website.

2) Prepare the memory server configuration. Create a global memory path or opt for a project-local memory directory. The following example uses a global path. You can adapt the path to your environment.

{
  "mcpServers": {
    "memory": {
      "command": "npx",
      "args": [
        "-y",
        "mcp-knowledge-graph",
        "--memory-path",
        "/Users/yourusername/.aim/"
      ]
    }
  }
}

Run the memory server

Use your MCP-enabled client to start the memory server with the configuration above. The client will launch the memory-graph tool and connect to the specified memory path. When the server starts, you will have access to: - A master database named default stored at the memory path - A work database named memory-work.jsonl - A personal database named memory-personal.jsonl - Additional databases as you create them with specific contexts (work, personal, health, etc.)

Available tools

aim_create_entities

Add new entities (people, projects, events) to the memory graph and associate initial observations.

aim_create_relations

Link two or more entities to establish relationships within the memory graph.

aim_add_observations

Attach factual statements or observations to an existing entity.

aim_search_nodes

Query the memory graph by keywords to locate entities or observations.

aim_read_graph

Output the current state of the entire memory graph for review.

aim_open_nodes

Retrieve specific entities by name for quick inspection.

aim_list_databases

List all available databases and indicate the current storage location.

aim_delete_entities

Remove entities from the memory graph.

aim_delete_observations

Remove specific facts associated with an entity.

aim_delete_relations

Delete relationships between entities.