Ever wish you could use a single, reliable memory for all your AI interactions, without sharing your data with multiple providers? `mem-agent-mcp` is a Model Context Protocol (MCP) server that lets you do just that. It's a local, private memory system that works with different LLMs.
**Tips for getting started:**
* **Keep your data local:** The memory system is based on Markdown files on your machine, giving you full control and privacy. You can even edit the memory manually.
* **One memory, multiple AIs:** Use the same memory instance with different LLM applications like Claude Desktop or LM Studio, ensuring consistent context across all your tools.
* **Import existing knowledge:** Use the provided connectors to pull in data from sources you already use, such as ChatGPT, Notion, and Google Docs, to quickly build out your memory.
*Important Privacy Note: While your memory is stored locally, any data you share with a remote LLM via the server will be visible to that LLM provider for the duration of the request.*
---
**Video Overview: A Deep Dive into MEM Agent**
This video provides a comprehensive tutorial on setting up and using MEM agent, demonstrating how you can give AI models like ChatGPT and Claude a private, infinite memory on your local machine.
You can watch the full video tutorial here: [https://www.youtube.com/watch?v=R4I_YaFYv3M](https://www.youtube.com/watch?v=R4I_YaFYv3M)
---
## Instructions:
### User Guide: Getting Started with `mem-agent-mcp`
This guide, based on the video "Give ChatGPT and Claude a private infinite memory on your local machine," will walk you through the process of setting up and using `mem-agent-mcp` to give your AI agents a private, local memory.
#### Section 1: Installation & Setup Guide
To get started, you'll need to set up the server and the necessary tools on your machine.
1. **Clone the GitHub Repository:** Find the official `mem-agent-mcp` GitHub repository at `https://github.com/firstbatchxyz/mem-agent-mcp` and clone it to your local machine. This will give you all the necessary files to run the server.
2. **Install LM Studio:** LM Studio is a required component for running the AI models locally. You can download it from `https://lmstudio.ai/`. Follow the instructions in the repository to install it on your system.
3. **Run the Agent:** Once you have the repository and LM Studio, run the `mem-agent-mcp` tool. This will initiate the setup process. During this step, you will install the 4 billion parameter MEM agent model locally.
4. **Configure Your Memory Directory:** You will be prompted to select a folder on your computer where your local memory will be stored. This is a critical step, as this folder will contain all your private data in Markdown format.
5. **Set Up the MCP JSON:** The tool will help you generate and configure the MCP JSON file. This file acts as the bridge between your memory agent and other applications.
6. **Integrate with Your AI Application:** Once the server is running, you can connect it to your favorite AI desktop app. For example, to integrate with Claude Desktop, which can be downloaded at `https://claude.ai/`, you will need to edit its configuration file as guided by the tool.
#### Section 2: How to Use Your New Memory
After a successful setup, you can begin interacting with your new private memory.
1. **Store Information with Natural Language:** You can ask your AI to store information for you using simple, natural language commands. For instance, you could say, "Remember that my project deadline is next Friday." The AI will process this request and save the information to your local memory files.
2. **Create Entities:** To organize information about specific topics, people, or projects, you can create "entities." This allows you to group related details, making it easier for the AI to retrieve and reference them later. For example, you can create an entity for "Company X" and store all relevant details under that heading.
3. **Filter Information:** You have the power to control what information the AI can access. This is particularly useful for protecting sensitive data. You can set up filters to prevent the AI from revealing specific details, such as confidential project names or personal information.
#### Section 3: Leveraging Connectors and Other Resources
To build a comprehensive memory, you can import data from sources you already use. `mem-agent-mcp` supports connectors for various software, including:
* ChatGPT
* Google Docs
* GitHub
* Notion
This allows you to chat with your own documents and repositories, all while keeping your data private and local.
For a complete visual walkthrough and more detailed instructions, please refer to the video tutorial: [https://www.youtube.com/watch?v=R4I_YaFYv3M](https://www.youtube.com/watch?v=R4I_YaFYv3M). If you are interested in diving deeper into AI development and learning about tools like Cursor, Claude, and Codex, you can join "The New Society," which is a community that offers exclusive content, weekly calls, and a one-on-one call with the video's creator. You can learn more about this community at `https://thenewsociety.co/classroom`.
http://googleusercontent.com/youtube_content/1