Mem0 MCP Server
Persistent long-term memory integration for AI agents via semantic search.
Data last scanned today · Reviewed today
Overview
This MCP server provides a template for integrating Mem0 into AI workflows, enabling agents to store, retrieve, and search information using semantic indexing. It is designed to demonstrate how to manage long-term state for AI assistants through established MCP patterns. The server includes specific tools to save new memories and perform semantic lookups. Users can deploy this implementation as a reference for building custom MCP servers or as a starting point for managing persistent context in AI-driven applications, utilizing PostgreSQL or Supabase for vector storage.
Our verdict
The Mem0 MCP Server carries a high-risk rating with a security score of 23/100. This is primarily due to the repository being community-maintained rather than official, and it has not received an editorial review for its implementation. Furthermore, the codebase has not seen updates in over 475 days. Potential users should approach this as a reference implementation or template for local development rather than a stable component for sensitive production environments.
- MIT licensed for flexible reuse
- Zero direct dependencies simplifies the supply-chain
- Includes detailed documentation in the README
- Provides functional tools for semantic memory retrieval
- Stale repository with no updates for 475 days
- High security risk due to lack of editorial review
- Community-maintained rather than provided by the official vendor
- Bus-factor risk with only one project contributor
Tools
| Tool | Description | Risk |
|---|---|---|
| save_memory | Store any information in long-term memory with semantic indexing | medium |
| get_all_memories | Retrieve all stored memories for comprehensive context | low |
| search_memories | Find relevant memories using semantic search | low |
Compatibility
| Client | Local | Docker | Remote | Read-only |
|---|---|---|---|---|
| Claude Desktop | ||||
| Cursor | ||||
| VS Code | ||||
| Windsurf | ||||
| ChatGPT |
Frequently asked questions
›What are the infrastructure requirements for this server?
You need Python 3.12 or newer, a PostgreSQL database or Supabase for vector storage, and an API key for your chosen LLM provider such as OpenAI, OpenRouter, or Ollama.
›What transport protocols are supported?
The server supports both SSE (Server-Sent Events) and stdio transport protocols.
›Which LLM providers can I use with this server?
You can configure the server to use OpenAI, OpenRouter, or Ollama as your LLM provider.
›Do I need to run this inside Docker?
Running the server as a container using Docker is recommended, but you can also run it directly using Python and uv.
›How do I configure connection settings for different MCP clients?
The server uses environment variables defined in a .env file to manage connection details, LLM settings, and database configurations. Some clients like Windsurf require specific adjustments, such as using 'serverUrl' instead of 'url' in the configuration.
Badge
Maintain this server? Add the live badge to your README.