memory

Overview

The memory-mcp-server is a Model Context Protocol (MCP) server that provides AI assistants with persistent, long-term memory across interactions. It allows AI-driven workflows to store, retrieve, and update information over time, making it possible for assistants to maintain context, preferences, and knowledge beyond a single conversation.

This server is especially useful for personalized assistants, long-running agents, and workflows that benefit from continuity and historical awareness.

Transport

streamable-http

Tools

• create_entities
• create_relations
• add_observations
• delete_entities
• delete_observations
• delete_relations
• read_graph
• search_nodes
• open_nodes

Key Capabilities

• Persistent context — Retain information across sessions and interactions.
• Personalization support — Remember user preferences, facts, and prior decisions.
• Contextual retrieval — Surface relevant memories at the right time during reasoning.
• Evolving knowledge — Update memory as new information is learned or corrected.
• Composable workflows — Combine memory with other MCP servers for richer reasoning.

How It Works

The memory MCP server runs as an MCP service that manages a persistent storage backend for memory entries. AI clients communicate with the server over the MCP protocol to store new information or retrieve relevant historical context as part of broader reasoning workflows.

The server handles indexing, retrieval, and updates, returning structured results that AI assistants can reason over and incorporate into responses or decisions. This abstraction allows AI workflows to maintain continuity without embedding storage or retrieval logic directly in the assistant.

By exposing memory through MCP, the server enables AI systems to develop continuity, personalization, and historical awareness — supporting more coherent and useful interactions over time.