Memory PostgreSQL

🚀 MCP Memory Server

The MCP Memory Server enables long-term memory functionality for AI assistants. It is based on the mem0 principle and uses PostgreSQL and pgvector for efficient vector similarity searches.

🚀 Quick Start

This server provides long - term memory capabilities for AI assistants. By leveraging the mem0 principle and using PostgreSQL along with pgvector, it enables efficient vector similarity searches.

✨ Features

• Conduct vector similarity searches using PostgreSQL and pgvector.
• Automatically generate embeddings using BERT.
• Offer RESTful APIs for memory operations.
• Provide semantic search capabilities.
• Support multiple types of memories (e.g., learning, experience).
• Enable tag - based memory retrieval.
• Implement memory confidence scoring.
• Use Server - Sent Events (SSE) for real - time updates.
• Be compatible with the Cursor MCP protocol.

📦 Installation

Prerequisites

1. Install PostgreSQL 14+ and enable the pgvector extension:

# In your PostgreSQL instance:
CREATE EXTENSION vector;

2. Node.js 16+

Setup

1. Install dependencies:

npm install

2. Configure environment variables: Copy .env.sample to .env and adjust the values:

cp .env.sample .env

Example .env configuration:

# Use username/password
DATABASE_URL="postgresql://username:password@localhost:5432/mcp_memory"
PORT=3333

# Use peer authentication for local development
DATABASE_URL="postgresql:///mcp_memory"
PORT=3333

3. Initialize the database:

npm run prisma:migrate

4. Start the server:

npm start

For a development environment with auto - reload:

npm run dev

💻 Usage Examples

Using with Cursor

Adding the MCP Server in Cursor

To add the memory server to Cursor, modify the MCP configuration file located at ~/.cursor/mcp.json. Add the following configuration to the mcpServers object:

{
  "mcpServers": {
    "memory": {
      "command": "node",
      "args": [
        "/path/to/your/memory/src/server.js"
      ]
    }
  }
}

Replace /path/to/your/memory with the actual installation path of the memory server.

For example, if you cloned the repository to /Users/username/workspace/memory, your configuration will be as follows:

{
  "mcpServers": {
    "memory": {
      "command": "node",
      "args": [
        "/Users/username/workspace/memory/src/server.js"
      ]
    }
  }
}

The server will start automatically when Cursor is launched. You can verify its status through the following steps:

1. Open Cursor.
2. The memory server will start automatically when Cursor launches.
3. To check the server status, visit http://localhost:3333/mcp/v1/health.

Available MCP Endpoints

SSE Connection

• Endpoint: GET /mcp/v1/sse
• Query Parameters:
  • subscribe: A comma - separated list of events to subscribe to (optional).
• Events:
  • connected: Sent upon initial connection.
  • memory.created: Sent when a new memory is created.
  • memory.updated: Sent when an existing memory is updated.

Memory Operations

1. Create Memory

POST /mcp/v1/memory
Content-Type: application/json

{
  "type": "learning",
  "content": {
    "topic": "Express.js",
    "details": "Express.js is a web application framework for Node.js"
  },
  "source": "Documentation",
  "tags": ["nodejs", "web - framework"],
  "confidence": 0.95
}

2. Search Memory

GET /mcp/v1/memory/search?query=web+frameworks&type=learning&tags=nodejs

3. List Memories

GET /mcp/v1/memory?type=learning&tags=nodejs,web - framework

Health Check

GET /mcp/v1/health

Response Format

All API responses follow the standard MCP format:

{
  "status": "success",
  "data": {
    // Response data
  }
}

Or in case of an error:

{
  "status": "error",
  "error": "Error message"
}

📚 Documentation

Memory Architecture

• id: A unique identifier.
• type: The type of memory (e.g., learning, experience).
• content: The actual memory content (in JSON format).
• source: The source of the memory.
• tags: A list of tags.
• created_at: The creation time.
• updated_at: The update time.