xgmem MCP Memory Server

What is xgmem MCP Memory Server?

xgmem is a TypeScript-based Model Context Protocol (MCP) server for project-specific and knowledge graph-based memory. It allows Claude, LLM agents, and other tools to store, retrieve, and manage entities, relations, and observations per project, with support for cross-project knowledge sharing and disk persistence.

How to use xgmem MCP Memory Server?

Add xgmem to your MCP client config (e.g., for Windsurf) using npx -y xgmem@latest. Install dependencies (npm install), build (npm run build), and run in development (npx ts-node index.ts) or production (npm start). For Docker, build the image and mount a volume to persist memory files. To save memories, call the save_project_observations tool via the MCP API with a project ID and observations.

Key features of xgmem MCP Memory Server

• Knowledge graph storage for entities, relations, and observations
• Full CRUD operations via MCP tools
• Disk persistence using a memory.json file per project
• Docker and TypeScript support for flexible deployment
• Configurable memory directory via MEMORY_DIR_PATH environment variable

Use cases of xgmem MCP Memory Server

• Agents and LLMs storing structured memory (entities, relations, observations) per project
• Cross-project knowledge sharing and migration between different agent sessions
• Scalable, disk-persistent, and queryable memory for agent ecosystems
• Saving and retrieving project-specific observations for context-aware AI interactions

FAQ from xgmem MCP Memory Server

How does xgmem persist memory?

xgmem saves memory to disk in a memory.json file. The storage directory can be changed by setting the MEMORY_DIR_PATH environment variable (default: /app/memories).

What tools does xgmem expose via MCP?

xgmem provides tools including save_project_observations, get_project_observations, add_graph_observations, create_entities, create_relations, delete_entities, delete_observations, delete_relations, read_graph, search_nodes, search_all_projects, open_nodes, and copy_memory.

How can I run xgmem with Docker?

Build the image with docker build -t xgmem-mcp-server . and run with a volume mount: docker run -v $(pwd)/memories:/app/memories xgmem-mcp-server. This persists all project memory files on your host.

What runtime does xgmem require?

xgmem requires Node.js and npm. It can be run directly on the command line via npx or after installing dependencies locally. It also works in Docker without requiring host Node.js installation.

Can xgmem share knowledge across projects?

Yes. xgmem is designed for cross-project knowledge sharing and migration. It stores memory per project but allows searching across all projects using the search_all_projects tool.