M

Memory MCP

Persistent memory across assistant sessions

Anthropic·No reviews yet
Repo
Share:

About

Persistent key-value memory across sessions. The assistant can store facts, preferences, and notes and retrieve them later without re-asking. Backed by a local JSON store — no server required.

Example tools

Illustrative tool names — the actual tool set is defined by the server at runtime.

create_task
Create a new task
list_tasks
List tasks with filters
update_task
Update fields of an existing task

Screenshots

No screenshots yet.

Frequently Asked Questions

Common questions about Memory MCP.

What is the Memory MCP used for?
The Memory MCP provides persistent key-value storage for assistants across sessions. It allows an assistant to store and retrieve facts, user preferences, or notes, eliminating the need to re-ask for information in subsequent interactions. This enhances continuity and personalization in assistant conversations.
How can I configure the Memory MCP in an MCP client?
The Memory MCP typically requires client-side configuration to specify the location of its local JSON store. Exact steps depend on your MCP client, such as Claude Desktop or Cursor, but generally involve adding the MCP by its slug 'memory' and ensuring proper file permissions for the JSON storage.
Does the Memory MCP require an API key or other authentication?
No, the Memory MCP does not require an API key, OAuth, or any other credentials for operation. It is designed to function without external authentication because it uses a local JSON store and does not interact with remote services.
What are the core capabilities of the Memory MCP?
The Memory MCP primarily exposes capabilities for storing and retrieving data as key-value pairs. This enables assistants to maintain context and remember information such as user preferences, past interactions, or specific details without needing to be reminded in each session.
What are the licensing costs for the Memory MCP?
The Memory MCP is open-source and typically incurs no direct licensing costs. Its implementation uses a local JSON store, so there are no associated infrastructure or service charges for its operation. Development and maintenance costs depend on internal resources.
What is a key limitation of the Memory MCP?
A key limitation of the Memory MCP is that its persistence is tied to a local JSON store. This makes it challenging to share memory across multiple assistant instances or different client devices. For distributed or synchronized memory, a different MCP or external service would be required.

Install Memory MCP

Claude Desktop

Add this to claude_desktop_config.json.

{
  "mcpServers": {
    "memory": {
      "command": "npx",
      "args": [
        "-y",
        "@modelcontextprotocol/server-memory"
      ]
    }
  }
}

Cursor

Add this to ~/.cursor/mcp.json.

{
  "mcpServers": {
    "memory": {
      "command": "npx",
      "args": [
        "-y",
        "@modelcontextprotocol/server-memory"
      ]
    }
  }
}

VS Code

Add this to your workspace settings.json.

{
  "mcp.servers": {
    "memory": {
      "type": "stdio",
      "command": "npx",
      "args": [
        "-y",
        "@modelcontextprotocol/server-memory"
      ]
    }
  }
}

Reviews

Sign in to leave a review.

No reviews yet. Be the first!

Discussion(0)

Sign in to join the discussion.

No comments yet.