Zmem

🚀 xgmem MCP Memory Server

xgmem is a TypeScript-based Model Context Protocol (MCP) server that provides project-specific and knowledge graph-based memory support for Claude, LLM agents, and other tools. It enables the storage, retrieval, and management of entities, relationships, and observations on a per-project basis, with a focus on flexibility and cross-project knowledge sharing.

🚀 Quick Start

✨ Features

• Knowledge graph storage supporting entities, relationships, and observations.
• CRUD operations via MCP tools.
• Storage to disk (memory.json).
• Support for Docker and TypeScript.

💻 Usage Scenarios

xgmem is suitable for:

• Agents and LLMs that need to store and retrieve structured memory, such as entities, relationships, and observations.
• Cross-project knowledge sharing and migration.
• An agent ecosystem with scalable, persistent disk storage, and query capabilities.

💻 Usage Examples

Basic Usage

MCP Configuration Example

Add the following to your MCP configuration file (e.g., windsurf):

"mcpServers": {
    "xgmem": {
      "command": "npx",
      "args": ["-y", "xgmem@latest"]
    }
  }
}

Install Dependencies

Run the following command to install all dependencies:

npm install

Build

Build using the following command:

npm run build

Run in Development Environment

Run the server in the development environment:

npx ts-node index.ts

Run in Production Environment

Run in the production environment:

npm start

Docker Deployment

Build the Docker image using the following command:

docker build -t xgmem-mcp-server .

Run the container and mount the persistent storage volume:

docker run -v $(pwd)/memories:/app/memories xgmem-mcp-server

In this way, the memory files of all projects will be persisted to the memories directory on the host.

Advanced Usage

How to Save Memory (MCP API)

To save observations for a project, you can call the save_project_observations tool of the MCP API: Example JSON:

{
  "name": "save_project_observations",
  "args": {
    "projectId": "demo-project",