myAI Memory Sync

An MCP server that synchronizes a customizable memory template across Claude interfaces (Claude Desktop, Claude Code, Windsurf).

Features

• Manage a standardized "myAI Memory" template with consistent formatting
• Support editing individual sections without affecting others
• Maintain the -~- prefix for preference items
• Synchronize template across multiple Claude interfaces automatically
• Support for preset profiles
• MCP-compliant server with both stdio and HTTP/SSE transport
• CLI for easy management of templates and presets

Installation

# Clone the repository
git clone https://github.com/yourusername/myai-memory-sync.git
cd myai-memory-sync

# Install dependencies
npm install

# Build the project
npm run build

Usage

MCP Server

Start the direct MCP server with stdio transport (for use with Claude Desktop):

./direct-mcp-server.sh

Or start with HTTP transport for remote connections:

./direct-http-server.sh

Template Format

The template follows a specific markdown format:

# myAI Memory

# User Information
## Use this information if you need to reference them directly
-~- Name: Default User
-~- Location: Default Location

# General Response Style
## Use this in every response
-~- Style: Concise and friendly

Each section starts with a # Section Title and can have an optional description with ## Description. Preferences are formatted as -~- Key: Value.

MCP Tools

The server exposes the following MCP tools:

Tool Name	Description
get_template	Get the full template
get_section	Get a specific section of the template
update_section	Update a section in the template
update_template	Update the entire template
list_presets	List available template presets
load_preset	Load a template preset
create_preset	Create a new preset from current template
sync_platforms	Sync template to platform(s)
list_platforms	List configured platforms

Claude Desktop Integration

To add this MCP server to Claude Desktop:

1. Open Claude Desktop
2. Go to Settings > Developer
3. Under "MCP Servers", add a new server:

   "myai-memory-sync": {
     "command": "/path/to/myai-memory-sync/direct-mcp-server.sh",
     "transport": "stdio"
   }
   

Replace /path/to/myai-memory-sync with the actual path to the repository.

Smithery Deployment

To deploy this MCP server to smithery:

1. Ensure all changes are committed to the repository

2. Build the project with npm run build

3. Push to the smithery repository:

   git push origin main
   

4. After deployment, update your MCP configuration to point to the correct file path:

   "myai-memory-sync": {
     "command": "node",
     "args": [
       "/path/to/myai-memory-sync/dist/direct-index.js"
     ],
     "transport": "stdio"
   }
   

Project Structure

myai-memory-sync/
├── data/              # Data storage
│   ├── presets/       # Preset templates
│   └── template.md    # Main template file
├── src/               # Source code
│   ├── services/      # Service implementations
│   ├── direct-index.ts  # Direct MCP server entry point (stdio)
│   ├── direct-mcp.ts    # Direct MCP server implementation
│   ├── direct-server.ts # Direct HTTP server implementation
│   ├── templateParser.ts  # Template parser
│   └── types.ts       # Type definitions
├── package.json
└── tsconfig.json

Development

# Run in development mode with auto-reload
npm run dev

# Build the project
npm run build

# Type check
npm run typecheck

# Lint the code
npm run lint

License

MIT