Memory MCP Server
A graph-based persistent memory server for AI agents to maintain long-term memory across sessions.
Preview
Problem
LLM chats lose memory between sessions, forcing developers to repeat preferences, architectural rules, and project context.
Solution
A local knowledge-graph MCP server that automatically stores entities, relations, and observations across conversations.
Challenge
Dynamically querying and updating graph relations without polluting the LLM prompt context.
Key Features
- Entity & Relation graph storage
- Continuous cross-session memory
- Search and retrieve facts
- Local JSON persistence
Tech Choices
Target User
Anyone using AI assistants across recurring, long-term engineering projects.
Impact
"Gave AI assistants true continuity and long-term memory across disjoint chat sessions."
How to Use
Install Memory Server
Add '@modelcontextprotocol/server-memory' to your MCP configuration.
Tell AI Facts
Say: 'Remember that our project uses Tailwind CSS v4 and PostgreSQL with Drizzle ORM'.
Recall in New Chats
In a new session, ask: 'What stack do we use for the database?' and the agent will retrieve it from memory.
Quick Details
- Pricing
- Free
- Platforms
- Node.js
- Updated
- 8/19/2026