Memory

Memory MCP Server provides a focused, local service for managing memories and daily notes through a modeled MCP interface. It enables AI agents to read, write, search, and summarize memories and notes, while offering task tracking and context awareness to improve decision making.

How to use

You connect to Memory MCP Server using an MCP client configured to load the memory server as a microservice. Once connected, you can read from long-term memory, add new memories, search through existing memories, manage daily notes, handle TODO lists, and retrieve a concise context summary for your AI agents. Use the documented tool endpoints to perform these actions and keep your memories organized by category and date.

How to install

Before you begin, make sure you have Node.js and npm installed on your system.

# Prerequisites: ensure Node.js and npm are installed on your system
node -v
npm -v

# Install the Memory MCP Server locally
cd /root/.openclaw/workspace/memory-mcp-server
npm install

# Start in development mode
npm start

# Or run directly with Node
node index.js

Additional notes on usage and integration

To integrate with your MCP client, load Memory MCP Server as a local MCP endpoint that exposes the available tools for memory and note management. The server exposes a set of actions including reading and writing to MEMORY.md, managing daily notes under memory/YYYY-MM-DD.md, and manipulating TODO items. You can start the server in development mode or run the main script directly for quicker iterations.

Configuration and runtime details

When run as a local MCP server, Memory MCP Server is typically started as a stdio process using Node. The explicit runtime command is shown in the integration example below, which you should include in your MCP configuration.

{
  "mcpServers": {
    "memory": {
      "type": "stdio",
      "name": "memory",
      "command": "node",
      "args": ["/root/.openclaw/workspace/memory-mcp-server/index.js"]
    }
  }
}

Available tools

get_memory

Read the contents of MEMORY.md (long-term memory) and retrieve stored memories.

add_memory

Add a new memory entry to MEMORY.md, optionally specifying a category.

search_memory

Search the contents of MEMORY.md using a query to find relevant memories.

get_daily_note

Read a daily note file at memory/YYYY-MM-DD.md or the latest entry for a specific date.

add_daily_note

Append an entry to a daily note for a given date.

list_daily_notes

List available daily notes with an optional limit.

get_todos

Retrieve TODO items for a specific date or default date.

add_todo

Create a new TODO item for a given date.

complete_todo

Mark a specified TODO item as complete for a given date.

get_context

Generate a concise summary of the current context for AI agents.