MCP Memory Server

A Model Context Protocol (MCP) server providing long-term memory storage for LLMs using a NoSQL database (TinyDB).

Features

• 🧠 Long-term memory storage across conversations
• 🔍 Search by keywords and tags
• 📊 Access tracking and statistics
• 🐳 Docker & Docker Compose ready
• 🌐 HTTP REST API for easy testing
• 💾 Persistent storage with TinyDB

Quick Start

Using Docker Compose (Recommended)

# Clone or create the repository
mkdir mcp-memory-server
cd mcp-memory-server

# Create all required files (see repository structure below)

# Start the server
docker-compose up -d

# Check logs
docker-compose logs -f

# Test the server
curl http://localhost:9393/health

Manual Installation

# Install dependencies
pip install -r requirements.txt

# Run the server
python server.py

API Endpoints

Health Check

curl http://localhost:9393/health

Store a Memory

curl -X POST http://localhost:9393/memories \
  -H "Content-Type: application/json" \
  -d '{
    "content": "User prefers dark mode",
    "tags": ["preference", "ui"],
    "metadata": {"importance": "high"}
  }'

Search Memories

# Search by keyword
curl -X POST http://localhost:9393/memories/search \
  -H "Content-Type: application/json" \
  -d '{"query": "dark mode", "limit": 5}'

# Search by tags
curl -X POST http://localhost:9393/memories/search \
  -H "Content-Type: application/json" \
  -d '{"tags": ["preference"], "limit": 10}'

Get a Specific Memory

curl http://localhost:9393/memories/1

Update a Memory

curl -X PUT http://localhost:9393/memories/1 \
  -H "Content-Type: application/json" \
  -d '{"content": "User prefers dark mode with blue accent"}'

Delete a Memory

curl -X DELETE http://localhost:9393/memories/1

Get Recent Memories

curl http://localhost:9393/memories/recent?limit=10

Get Frequently Accessed Memories

curl http://localhost:9393/memories/frequent?limit=10

Get Statistics

curl http://localhost:9393/stats

Get All Tags

curl http://localhost:9393/tags

Repository Structure

mcp-memory-server/
├── server.py              # Main server code
├── requirements.txt       # Python dependencies
├── Dockerfile            # Docker image definition
├── docker-compose.yml    # Docker Compose configuration
├── README.md            # This file
├── .gitignore           # Git ignore rules
└── data/                # Persistent data directory
    └── memory_bank.json # TinyDB database (created automatically)

Configuration

The server runs on port 9393 by default. To change:

1. Update docker-compose.yml ports mapping
2. Update the port in server.py (line with TCPSite)

MCP Integration

To use with Claude Desktop or other MCP clients:

{
  "mcpServers": {
    "memory": {
      "command": "docker",
      "args": ["run", "-i", "--rm", "-v", "./data:/data", "mcp-memory-server"]
    }
  }
}

Development

# Build the Docker image
docker build -t mcp-memory-server .

# Run with custom data directory
docker run -p 9393:9393 -v $(pwd)/custom-data:/data mcp-memory-server

# View logs
docker-compose logs -f

# Stop the server
docker-compose down

Data Persistence

The database is stored in /data/memory_bank.json inside the container, which is mounted to ./data/ on your host machine. This ensures memories persist across container restarts.

License

MIT