myAI Memory Sync

Synchronizes memory templates across different Claude interfaces.

Tired of repeating yourself to Claude every time you start a new chat?myAI Memory Sync is a game-changing MCP tool that seamlessly synchronizes your preferences, personal details, and code standards across ALL your Claude interfaces! Just update once, and your changes instantly appear everywhere - from Claude Desktop to Claude Code, Windsurf, and Claude.ai web. With our cutting-edge caching system, memory-related queries are up to 2000x faster! Stop wasting tokens on repetitive instructions and enjoy a truly personalized AI experience.

How myAImemory-mcp Compares to Other Memory Tools

While several excellent memory tools exist for AI systems, myAImemory-mcp serves a specific purpose as a Model Context Protocol (MCP) tool:

- Privacy-First: All data remains on your device, no personal information sent to external servers
- Performance: Leverages Claude's caching capabilities for dramatic speed improvements
- Simplicity: Natural language updates to your preferences across all Claude interfaces
- MCP Integration: Purpose-built as a Claude MCP for seamless integration

# Clone repository git clone https://github.com/Jktfe/myaimemory-mcp.git cd myaimemory-mcp # Install dependencies npm install # Build TypeScript code npm run build # Start MCP server (with stdio transport) npm start # Or start with HTTP transport npm run start:http

The unified server script supports multiple options:

# Start with stdio transport (default) ./start-server.sh # Start with HTTP transport ./start-server.sh --http # Start with HTTP transport on custom port ./start-server.sh --http --port=8080 # Start with direct implementation (no SDK) ./start-server.sh --direct # Start with direct implementation and HTTP transport ./start-server.sh --direct --http # Enable debug mode ./start-server.sh --debug

🔄 Direct Sync Method (Simple Alternative)

For a simpler approach that doesn't require running an MCP server, you can use the unified CLI:

# One-time sync of all memory files npm run sync # Or for emergency sync (fixes permissions) npm run sync:emergency

- Read from your "myAI Master.md" file
- Update all CLAUDE.md files in your projects
- Update your Windsurf memory settings
- All without storing sensitive information in the git repository

- The "myAI Master.md" file with your personal information is excluded from git tracking
- All CLAUDE.md files are also excluded to protect your privacy
- Use the included.gitignoreto ensure sensitive files remain private

You can interact with myAI Memory using these natural language patterns:

Note: To perform a full sync across all platforms, use the command line:node sync-memory.js

You: Use myAI Memory to remember I prefer TypeScript over JavaScript Claude: ✅ Added to your Coding Preferences! I'll remember you prefer TypeScript over JavaScript.

# Start with stdio transport (default) myai # Start with HTTP transport myai server --transport http # Process memory commands myai remember "I prefer dark mode" # Sync across platforms myai sync

git clone https://github.com/Jktfe/myaimemory-mcp.git cd myaimemory-mcp npm install npm run build npm start # Start with stdio transport # or npm run start:http # Start with HTTP transport

docker build -t myai-memory-sync . docker run -v myai-memory:/app/data -p 3000:3000 myai-memory-sync

Add this to yourclaude_desktop_config.json:

{ "mcpServers": { "myai-memory-sync": { "command": "npx", "args": [ "-y", "myai" ], "env": { "TEMPLATE_PATH": "/path/to/custom/template.md", "ENABLE_ANTHROPIC": "true", "ANTHROPIC_API_KEY": "your-api-key-here" } } } }

- VisitSmithery.ai
- Add the myAI Memory Sync MCP:

@Jktfe/myaimemory-mcp

{ "mcp": { "servers": { "myai-memory-sync": { "command": "npx", "args": [ "-y", "myai" ] } } } }

# Using npm scripts: npm run start:http # Using the unified CLI: myai server --transport http # Using the shell script with custom port: ./start-server.sh --http --port=8080 # Using environment variable: PORT=8080 npm run start:http

Create a.envfile with the following options:

# Basic configuration DEBUG=true # Enable debug logging TEMPLATE_PATH=./data/template.md # Custom template location PORT=3000 # Port for HTTP transport (default: 3000) USE_DIRECT=true # Use direct implementation (no SDK) # Platform-specific paths WINDSURF_MEMORY_PATH=~/.codeium/windsurf/memories/global_rules.md CLAUDE_PROJECTS_PATH=~/CascadeProjects # Performance optimization ENABLE_ANTHROPIC=true # Enable Anthropic API integration ANTHROPIC_API_KEY=your-api-key # Your Anthropic API key ENABLE_PROMPT_CACHE=true # Enable prompt caching system CACHE_TTL=300000 # Cache TTL in milliseconds (5 minutes) # Claude web sync (optional) CLAUDE_WEB_SYNC_ENABLED=false # Enable Claude.ai web synchronization CLAUDE_WEB_EMAIL=you@email.com # Your Claude.ai email CLAUDE_WEB_HEADLESS=true # Run browser in headless mode

For best results, add this to your Claude system prompt:

Memory Integration Instructions: When you receive a command that starts with "use myAI Memory to", you should: 1. Process the rest of the instruction as a memory management command 2. Try to determine the appropriate section to update based on the content 3. Use the myAI Memory Sync MCP to update your memory 4. Confirm the update with a brief acknowledgment For example: "use myAI Memory to remember I prefer dark mode" → Update the preferences section with dark mode preference When asked questions about preferences or personal information, first check your memory via the myAI Memory Sync MCP. Always reference information from memory rather than making assumptions.

- 🔄Cross-Platform Synchronization: Update once, syncs everywhere
- ⚡Lightning-Fast Recall: Caching system with up to 2000x performance boost
- 🗣️Natural Language Interface: Just talk naturally to update your preferences
- 🧩Multiple Persona Profiles: Switch between different presets with ease
- 🔐Security-Focused: Local storage with .gitignore protection
- 🛠️Developer-Friendly: Full TypeScript implementation with comprehensive API

myAI Memory Sync uses a modular architecture with these key components:

- Template Parser: Bidirectional conversion between structured memory objects and markdown
- Template Storage: Persistent storage with in-memory and file-system caching
- Platform Synchronizers: Implements thePlatformSyncerinterface for each target platform
- Natural Language Processor: Extracts structured data from natural language memory commands
- Memory Cache Service: Optimizes performance with multi-level caching strategies

- ClaudeCodeSyncer: Updates CLAUDE.md files across all repositories
- WindsurfSyncer: Manages global_rules.md in Windsurf environment
- ClaudeWebSyncer: Optional Puppeteer-based synchronization with Claude.ai web interface

- Pattern-Based Extraction: Converts natural language to structured key-value pairs
- Section Detection Algorithm: Automatically determines appropriate section for new memories
- Memory Template Format: Markdown-based structure with sections, descriptions, and key-value items
- Context Preservation: Updates memory sections while preserving other template content

- Multi-Level Caching: In-memory caching at both template and section levels
- TTL-Based Cache Management: Configurable Time-To-Live for cached content
- Pre-Warming: Cache pre-population after template updates
- Optional Anthropic API Integration: Accelerates memory-related queries up to 2000x

- Local-First Architecture: All data remains on your device
- Gitignore Management: Automatically adds CLAUDE.md to .gitignore in all repositories
- File Permission Handling: Fixes permissions issues for maximum compatibility
- Encrypted Storage: Compatible with encrypted file systems

The system uses a structured markdown format to organize your preferences:

# myAI Memory # User Information ## Use this information if you need to reference them directly -~- Name: Your Name -~- Location: Your Location -~- Likes: Reading, Hiking, Technology # General Response Style ## Use this in every response -~- Style: Friendly and concise -~- Use UK English Spellings: true -~- Include emojis when appropriate: true # Coding Preferences ## General Preference when responding to coding questions -~- I prefer TypeScript over JavaScript -~- Show step-by-step explanations

interface MemoryTemplate { sections: TemplateSection[]; } interface TemplateSection { title: string; description: string; items: TemplateItem[]; } interface TemplateItem { key: string; value: string; }

interface PlatformSyncer { sync(templateContent: string): Promise<SyncStatus>; } type PlatformType = 'claude-web' | 'claude-code' | 'windsurf' | 'master'; interface SyncStatus { platform: PlatformType; success: boolean; message: string; }

The myAI Memory Sync tool implements the Model Context Protocol (MCP) with the following functions:

Users can interact with the system through natural language commands:

You: Use myAI Memory to remember I prefer TypeScript over JavaScript Claude: ✅ Added to your Coding Preferences! I'll remember you prefer TypeScript over JavaScript. You: Use myAI Memory to load preset developer Claude: ✅ Loaded developer preset! I'll now use your developer preferences.

Switch between different personas easily:

You: Use myAI Memory to list presets Claude: Available presets: personal, work, developer You: Use myAI Memory to load preset developer Claude: ✅ Loaded developer preset!

When you need to fix synchronization issues across all platforms:

# Sync everything immediately ./emergency-sync.sh

# View all available commands node dist/cli.js --help # Process memory commands directly node dist/cli.js --remember "remember I prefer dark mode" # Start HTTP server for SSE transport npm run start:http # Start stdio server for MCP transport npm run start

# Run in development mode with auto-reload npm run dev # Run in development mode with HTTP server npm run dev:http # Watch TypeScript compilation npm run build:watch # Run tests npm test # Run specific test npm test -- -t "platformSync" # Lint code npm run lint # Type check without emitting files npm run typecheck

Our caching system delivers incredible performance improvements:

- All data remains locally on your device
- CLAUDE.md files are automatically added to .gitignore
- No data is sent to external servers (except when using the optional Anthropic API integration)
- Works with encrypted file systems for maximum security

- Check file permissions withls -la CLAUDE.md
- Try emergency sync with./emergency-sync.sh
- Verify platform paths in your.envfile

- Ensure MCP server is running withps aux | grep myai-memory
- Check Claude Desktop logs for MCP errors
- Verify your Claude Desktop configuration file

- Clear cache withnode dist/cli.js --clear-cache
- Verify Anthropic API key is correctly set
- Check memory file integrity withnode dist/cli.js --validate

- Make sure to use exactly one of the supported command patterns (see Supported Natural Language Commands section)
- If Claude doesn't recognize your command, try a different pattern
- For syncing across all platforms, use the direct script:node sync-memory.js

If you're experiencing issues with natural language commands or the MCP server:

# Direct sync approach (most reliable) cd /path/to/myAImemory node sync-memory.js # Alternative emergency sync (if permissions need fixing) cd /path/to/myAImemory ./safe-memory.sh sync

These methods directly read from your master file and update all platforms without relying on the MCP server or natural language processing.

- Linux/macOS:~/.local/share/myai-memory/logs/
- Windows:%APPDATA%\myai-memory\logs\

Contributions are welcome! Please feel free to submit a Pull Request.
- Fork the repository
- Create your feature branch (git checkout -b feature/amazing-feature)
- Commit your changes (git commit -m 'Add some amazing feature')
- Push to the branch (git push origin feature/amazing-feature)
- Open a Pull Request

We follow a standard Git workflow and CI process:

- All PRs require passing tests and linting
- New features should include tests
- Major changes should update documentation
- Follow existing code style and patterns

For more detailed documentation, see theWiki.

# Generate API documentation npm run docs

This project is licensed under the MIT License - see theLICENSEfile for details.

Project Link:https://github.com/Jktfe/myaimemory-mcp

AI conversation memory that works everywhere — save and recall across Claude, ChatGPT, Gemini, Cursor, and all MCP-compatible platforms. 11 tools including shared community memories.