MCP Catalogs
Homemem0-mcp screenshot

mem0-mcp

by pinkpixel-dev·95·Score 49

An MCP server providing persistent memory for AI agents using Mem0.ai with cloud, Supabase, or local storage options.

ai-llmdeveloper-toolsdatabase
13
Forks
1
Open issues
1 mo ago
Last commit
2d ago
Indexed

Overview

The mem0-mcp server enables AI agents to store and retrieve memories across sessions through the Model Context Protocol. It provides three storage modes: cloud storage (recommended for production), Supabase (for self-hosting), and local storage (for development/testing). The server exposes three main tools: add_memory, search_memory, and delete_memory, with extensive configuration options for each. Integration is straightforward with multiple installation methods including global npm install, npx execution, or direct from a cloned repository.

Try asking AI

After installing, here are 6 things you can ask your AI assistant:

you:Enabling long-term memory in conversational AI agents
you:Personalizing AI interactions based on user history
you:Building AI assistants that remember user preferences across sessions
you:What storage options are available?
you:Do I need a database for local storage?
you:What API keys are required?

When to choose this

Choose mem0-mcp when you need persistent memory for AI agents without building a custom memory system, especially if you're already using or comfortable with Supabase or want cloud-hosted memory solutions.

When NOT to choose this

Avoid mem0-mcp if you need full control over your data storage infrastructure, have strict data residency requirements, or prefer a self-contained solution without external API dependencies.

Tools this server exposes

3 tools extracted from the README
  • add_memory

    Stores a piece of text content as a memory associated with a specific userId.

  • search_memory

    Searches stored memories based on a natural language query for a specific userId.

  • delete_memory

    Deletes a specific memory from storage by its ID.

Comparable tools

semantic-memory-mcpvector-store-mcpmemory-dbsemantic-kernel-memory

Installation

Installation

Via Smithery

npx -y @smithery/cli install @pinkpixel-dev/mem0-mcp-server --client claude

Global Installation

npm install -g @pinkpixel/mem0-mcp

Configure Claude Desktop

Add this to your Claude Desktop config.json:

**Cloud Storage:**

{
  "mcpServers": {
    "mem0-mcp": {
      "command": "mem0-mcp",
      "env": {
        "MEM0_API_KEY": "YOUR_MEM0_API_KEY",
        "DEFAULT_USER_ID": "user123"
      },
      "alwaysAllow": [
        "add_memory",
        "search_memory",
        "delete_memory"
      ]
    }
  }
}

**Supabase Storage:**

{
  "mcpServers": {
    "mem0-mcp": {
      "command": "mem0-mcp",
      "env": {
        "SUPABASE_URL": "YOUR_SUPABASE_URL",
        "SUPABASE_KEY": "YOUR_SUPABASE_KEY",
        "OPENAI_API_KEY": "YOUR_OPENAI_API_KEY"
      },
      "alwaysAllow": [
        "add_memory",
        "search_memory",
        "delete_memory"
      ]
    }
  }
}

FAQ

What storage options are available?
Three storage modes: cloud (recommended for production), Supabase (for self-hosting), and local (development/testing only).
Do I need a database for local storage?
No, local storage uses an in-memory vector database. Data is lost on server restart unless configured for persistence.
What API keys are required?
Cloud mode requires MEM0_API_KEY. Supabase mode requires SUPABASE_URL, SUPABASE_KEY, and OPENAI_API_KEY. Local mode requires OPENAI_API_KEY.

Compare mem0-mcp with

GitHub →

Last updated · Auto-generated from public README + GitHub signals.