Memory MCP server

Memory MCP server is the Memory MCP server is a reference Model Context Protocol server that gives AI assistants persistent memory in a local knowledge graph.

The Memory MCP server is one of the reference servers kept in the Model Context Protocol project's servers repository. It stores what an assistant learns as a knowledge graph of entities, relations between them and short observations attached to each entity, saved to a local file. An MCP client such as Claude Desktop, VS Code or Cline calls its tools to create entities and relations, add or delete observations, read the whole graph, search nodes by a query and open specific nodes by name. Because the graph stays on disk, an assistant can recall details about a person, project or preference in later chats. It is published on npm as @modelcontextprotocol/server-memory, runs with npx or Docker, and the storage file is set with the MEMORY_FILE_PATH variable. It is free and open source.

Memory MCP server at a glance

Best for: Software engineer
Runs on: VS Code, CLI, self-hosted, open source
Pricing: Free and open source
Released: November 2024
Category: AI coding tool
Website: github.com/modelcontextprotocol/servers

Key features

Knowledge graph of entities, relations and observations
Tools to create, update and delete graph items
Search nodes by query and open nodes by name
Memory saved to a local file between chats
Runs with npx or Docker
Custom storage path through MEMORY_FILE_PATH

Use it for

Let an assistant remember preferences across chats
Keep project facts available to a coding agent between sessions
Record people and how they relate for later recall

Companion to

Tools whose articles name Memory MCP server as a companion, each with its job for them.

Cline — Memory MCP server: Memory

Alternatives to Memory MCP server

Codebase Memory MCP — High-speed indexing and persistent knowledge graphs for AI coding agents.
ContextMCP — Maintains fresh, accurate context for AI agents using AST-aware chunking.
IAI Memory Engine — Provides persistent, local, and encrypted memory for AI agents without APIs.
Official MCP Servers — Find trusted official MCP servers without unofficial forks or abandoned projects.
MCP Server Finder — It provides a comprehensive directory with detailed profiles and compatibility matrices.
Home Assistant MCP — Connects AI assistants to Home Assistant using natural language.
MCP Toolbox for Databases — Connects AI agents to diverse enterprise databases with built-in security and observability.
claude-mem — It shares persistent, searchable agent memory across sessions and devices.
Grafana MCP — Connects AI agents to Grafana data and dashboard configurations.
Spec Workflow MCP — It brings sequential spec creation and real-time tracking to your AI workflow.
GitHub MCP Server — Connects AI tools directly to GitHub for natural language code management.
AWS MCP Servers — Provides real-time access to official AWS documentation and infrastructure tools.

All Memory MCP server alternatives