MemGPT MCP Server: Integrated Memory System for LLMs

Overview: What is MemGPT MCP Server?

MemGPT, an MCP-based server developed in TypeScript, provides a robust memory system that enhances interaction between Artificial Intelligence Language Models (LLMs) and their users. By implementing the Model Context Protocol (MCP), it seamlessly integrates with various AI applications to maintain conversation history, switch models, and providers without disrupting user experience or conversation continuity.

🔧 Core Features & MCP Capabilities

MemGPT MCP Server supports multiple LLM providers, enabling users to seamlessly switch between different platforms like OpenAI, Anthropic, OpenRouter, and Ollama. The server offers tools for various tasks such as sending messages, retrieving conversation history, clearing memories, switching providers, and models.

Chat Tool

The chat tool sends a message to the currently selected LLM provider with options for multiple providers:

• Parameters: Takes a message parameter.
• Provider Support: Supports OpenAI, Anthropic, OpenRouter, and Ollama.

Memory Management Tools

Retrieve Conversation History (get_memory)

Retrieves conversation history from stored memories. Offers control over the number of memories to retrieve through an optional limit parameter:

• Limit Parameter: Optional limit allows for limiting memory retrieval.
• Unlimited Retrieval: Use { "limit": null } for retrieving all stored memories.
• Order and Timestamps: Returns memories in a chronological order with timestamps.

Clear Memory (clear_memory)

Completely clears the conversation history, effectively removing all stored memories.

Switch Provider (use_provider)

Switches between different LLM providers while persisting provider selection:

• Supported Providers: OpenAI, Anthropic, OpenRouter, and Ollama.

Switch Model (use_model)

Switch to a different model within the current provider for enhanced functionality. Provides support for various models across different providers:

• Anthropic Claude Models:

  • Claude 3 Series:
    • claude-3-haiku: Fast response times suitable for customer support and content moderation.
    • claude-3-sonnet: Balanced performance for general-purpose use.
    • claude-3-opus: Advanced model ideal for complex reasoning and high-performance tasks.
  • Claude 3.5 Series:
    • claude-3.5-haiku: Enhanced speed and cost-effectiveness.
    • claude-3.5-sonnet: Superior performance with capabilities in computer interaction.

• OpenAI:

  • Models: 'gpt-4o', 'gpt-4o-mini', 'gpt-4-turbo'.

• OpenRouter: Any model specified as 'provider/model' (e.g., openai/gpt-4, anthropic/claude-2).

• Ollama: Any locally available model like 'llama2', 'codellama'.

All these selections persist across sessions.

⚙️ MCP Architecture & Protocol Implementation

MemGPT is built on TypeScript, focusing on implementing the Model Context Protocol (MCP) to ensure seamless communication between the server and various AI applications. The core architecture includes:

• Intercommunication: Standardized protocol over stdio for efficient data exchange.
• Flexibility: Multiple LLM providers handled through a unified interface.

MCP Protocol Flow Diagram

The following diagram illustrates the flow of signals and responses between an AI application, MemGPT server, and backend tools or data sources:

graph TD
    A[AI Application] -->|MCP Client| B[MCP Protocol]
    B --> C[MCP Server]
    C --> D[Data Source/Tool]
    style A fill:#e1f5fe
    style C fill:#f3e5f5
    style D fill:#e8f5e8

Realistic AI Workflow Scenarios

Case Study 1: Customer Support with MemGPT MCP Server

An integration is established between the MemGPT MCP Server and a support chatbot. The flow involves initiating a conversation, retrieving previous exchanges, and switching to a more advanced model for better resolution:

• Initialization: User starts a chat session.
• Memory Retriever (get_memory): Pulls up recent issue details.
• Model Switching (use_model): Upgrades model to claude-3-opus.
• Response Generation (chat): Generates detailed solution.

Case Study 2: Content Creation with Anthropic’s Claude

A writer uses Claude Desktop, incorporating MemGPT MCP Server for a smoother experience:

• Setup: Configures Claude to use MemGPT MCP Server.
• Conversation History (get_memory): Retrieves recent drafts and ideas from the writing session.
• Model Adjustment (use_model): Enhances model to claude-3.5-sonnet.
• Content Generation (chat): Writes new content seamlessly.

🚀 Getting Started with Installation

To install MemGPT MCP Server, follow these steps:

1. Install Dependencies: Run the following command in your terminal.

   npm install
   

2. Build the Server:

   npm run build
   

3. Development with Auto-Rebuild: For real-time development and server rebuilding, use:

   npm run watch
   

4. Configure for Use with Claude Desktop:

   • MacOS: ~/Library/Application Support/Claude/claude_desktop_config.json
   • Windows: %APPDATA%/Claude/claude_desktop_config.json

5. Add the Following Configuration:

   {
     "mcpServers": {
       "letta-memgpt": {
         "command": "/path/to/memgpt-server/build/index.js",
         "env": {
           "OPENAI_API_KEY": "your-openai-key",
           "ANTHROPIC_API_KEY": "your-anthropic-key",
           "OPENROUTER_API_KEY": "your-openrouter-key"
         }
       }
     }
   }
   

6. Environment Variables:

   • OPENAI_API_KEY
   • ANTHROPIC_API_KEY
   • OPENROUTER_API_KEY

💡 Key Use Cases in AI Workflows

MemGPT MCP Server is ideal for scenarios requiring persistent conversation history and model switching flexibility, such as:

• Customer Support: Maintaining context across multiple customer interactions.
• Content Creation: Managing and reusing previous content drafts during the writing process.

These workflows leverage the server’s robust capabilities to provide a seamless AI interaction experience.

🔌 Integration with MCP Clients

MemGPT is compatible with several MCP clients, enhancing their functionality by integrating multiple LLM providers and models:

MCP Client	Resources	Tools	Prompts	Status
Claude Desktop	✅	✅	✅	Full Support
Continue	✅	✅	✅	Full Support
Cursor	❌	✅	❌	Tools Only

📊 Performance & Compatibility Matrix

The performance and compatibility of MemGPT MCP Server are optimized for:

• Cross-Platform Operation: Works seamlessly across different operating systems.
• Multiple Model Integration: Supports various models, ensuring robustness in diverse use cases.

🛠️ Advanced Configuration & Security

Advanced configuration and security measures include:

• Environmental Setup: Securely manage API keys via environment variables or secure storage solutions.
• Authentication & Authorization: Implementing proper authentication methods to protect data privacy.
• CORS Policy Management: Securing APIs from unauthorized access with appropriate CORS policies.

❓ Frequently Asked Questions (FAQ)

1. Q: What is the difference between MemGPT and other MCP servers?

   • A: MemGPT offers extensive support for multiple LLM providers, models, and enhanced memory management features.

2. Q: How does MemGPT handle security with API keys?

   • A: Securely store and manage API keys through environment variables or secure storage solutions to protect data privacy.

3. Q: Can I use MemGPT with Claude Desktop on Windows?

   • A: Yes, integration is supported on both macOS and Windows platforms, as demonstrated in the configuration examples.

4. Q: What tools does MemGPT support besides LLMs?

   • A: Besides various LLM providers, it supports tools for memory management, model switching, and more.

5. Q: Can I use MemGPT with models not listed in the README?

   • A: Yes, MemGPT is flexible enough to accommodate custom or third-party models.

👨‍💻 Development & Contribution Guidelines

Contribution to MemGPT MCP Server involves understanding its architecture and following best practices for TypeScript development. Key steps include:

• Pull Requests: Fork the repository and create pull requests for bug fixes or feature enhancements.
• Testing: Ensure comprehensive testing of changes before submission.

🌐 MCP Ecosystem & Resources

For further information on MCP, visit the official Model Context Protocol documentation at ModelContextProtocol.org.

---

By providing robust integration capabilities and advanced features like memory management and model switching, MemGPT MCP Server is a valuable addition to AI application development, ensuring seamless user experience across diverse platforms.