Mem0

OpenAI just added memory across your chats across your openAI account. But wouldn't it be awesome to have general AI memory across all your interactions with any and all AI tools, IDEs, chatbots.... Now if it supports MCP you can with https://mem0.ai/ Give Claude desktop memory. Give cursor or windsurf memory across sessions or different projects.

python

GitHub Stars

python

Language

4 months ago

First Indexed

3 weeks ago

Catalog Refreshed

Documentation & install

Readme and setup notes from the catalogue, plus a client-ready config you can copy for your MCP host.

Installation

Add the following to your MCP client configuration file.

Configuration

View docs

{
  "mcpServers": {
    "ryaker-mcp-mem0-general": {
      "command": "uvx",
      "args": [
        "git+https://github.com/ryaker/mcp-mem0-general.git",
        "mcp-mem0-general"
      ],
      "env": {
        "MEM0_API_KEY": "your-mem0-api-key-here"
      }
    }
  }
}

You can run the Mem0 MCP Server to give AI assistants access to Mem0 memory, enabling storing, retrieving, and querying memories that power persistent context and knowledge graphs. This MCP bridge lets your AI agents seamlessly work with Mem0’s memory system to enhance conversations and tooling.

How to use

Connect your MCP client to the Mem0 MCP Server to store, retrieve, and search memories across episodic, semantic, and procedural types. Use the server to maintain short-term context during conversations and apply selective memory patterns to control what gets stored. Create knowledge graphs from memories to reveal relationships between entities and concepts. Start by ensuring your client is configured with the proper API key and points to the Mem0 MCP endpoint provided by the installation method you choose.

How to install

Prerequisites: ensure you have a working Python or Node environment and a command that can run MCP servers. You also need access to a Mem0 API key for authentication.

Option A — Use the direct run method via uvx (recommended). Run the server directly from its MCP entry point without cloning the repository.

Step 1: Set your Mem0 API key in your environment.

Step 2: Start the server with uvx using the MCP package and entry name.

export MEM0_API_KEY="your-mem0-api-key-here"

uvx git+https://github.com/ryaker/mcp-mem0-general.git mcp-mem0-general

Additional configuration and notes

If you encounter a warning about Neo4j libraries, you can safely ignore it when using the managed Mem0.ai platform. If you self-host Mem0 with Neo4j, install the required libraries (langchain-neo4j, neo4j) manually to enable graph features.

Plan for memory data management by configuring memory types and patterns from your MCP client. Short-term memories handle recent conversations and working context, while long-term memories encompass episodic, semantic, and procedural data. Use the graph features to map relationships between memories and entities.

Security and environment considerations

Protect your Mem0 API key and restrict access to the MCP endpoints. Store configuration in a secure place and set environment variables only in trusted environments.

Available tools

mem0_add_memory

Store a new memory into Mem0 with optional metadata and memory type.

mem0_search_memory

Query memories using semantic similarity to find relevant past memories.

mem0_get_memory_by_id

Retrieve a memory by its unique memory ID.

mem0_create_knowledge_graph

Create and query knowledge graph relationships between memory entities.

mem0_selective_memory

Apply include/exclude patterns to filter text before storing memories.

Built by

VeilStrat

AI signals for GTM teams