D.E.R.E.K Memory MCP Server

Share AI project specifications via Model Context Protocol

A simple MCP server that exposes your project's .kiro memory files (requirements, design docs, tasks) as resources for AI assistants, with tools to generate and serve beautiful HTML documentation.

📦 Installation

Option 1: Using npx (Recommended)

No installation needed! Just add to your MCP config:

{
  "mcpServers": {
    "derek-memory": {
      "command": "npx",
      "args": ["-y", "derek-memory-mcp"]
    }
  }
}

Option 2: Global Installation

npm install -g derek-memory-mcp

Then configure:

{
  "mcpServers": {
    "derek-memory": {
      "command": "derek-memory-mcp"
    }
  }
}

Option 3: Local Development

Clone and use the local script:

{
  "mcpServers": {
    "derek-memory": {
      "command": "node",
      "args": ["/absolute/path/to/derek-memory-mcp-node/index.js"]
    }
  }
}

✨ Quick Setup

That's it! The server auto-detects your .kiro folder in your current project.

📋 What It Provides

Resources (Read-Only)

• Global memory: PROJECT.md, CONTEXT.md, ARCHITECTURE.md, etc.
• Feature memory: requirements.md, design.md, tasks.md, notes.md

Tools (Actions)

1. share_feature_memory - Generate beautiful HTML documentation
2. serve_html - Start HTTP server to view HTML in browser

🎯 Usage Examples

Read Project Specs

"What resources are available from derek-memory?"
"Read the PROJECT.md resource"
"Show me the requirements for skill-manager feature"

Generate HTML Documentation

"Generate a shareable page for skill-manager"

Creates a beautiful HTML file with:

• ✅ Tabbed interface (Requirements, Design, Tasks, Notes)
• ✅ Markdown rendering with syntax highlighting
• ✅ Professional styling
• ✅ Mobile-responsive design

View in Browser

"Start an HTTP server to view it"

Opens at http://localhost:8080/ - perfect for sharing with your team!

📁 Project Structure Required

Your project needs a .kiro folder:

your-project/
├── .kiro/
│   ├── resources/          # Global memory
│   │   ├── PROJECT.md
│   │   ├── CONTEXT.md
│   │   └── ...
│   └── features/           # Feature memory
│       └── feature-name/
│           ├── requirements.md
│           ├── design.md
│           ├── tasks.md
│           └── notes.md

🔒 Security

Automatically sanitizes sensitive data before sharing:

• API keys → [REDACTED]
• Passwords → [REDACTED]
• Tokens → [REDACTED]
• Credentials → [REDACTED]

🚀 Features

✅ Auto-detection - Finds .kiro folder automatically (up to 5 parent levels)
✅ Universal config - Same config works for all projects
✅ Beautiful HTML - Professional documentation with tabs and markdown
✅ HTTP server - Built-in server for viewing HTML
✅ Sanitization - Automatic removal of sensitive data
✅ No installation - Just point to the script

🛠️ Development

Install Dependencies

npm install

Run Locally

node index.js

Test

Open a project with .kiro folder and ask your AI to list resources.

📖 Configuration Locations

Claude Desktop

macOS: ~/Library/Application Support/Claude/claude_desktop_config.json
Windows: %APPDATA%\Claude\claude_desktop_config.json
Linux: ~/.config/Claude/claude_desktop_config.json

Cline (VS Code)

Settings → Extensions → Cline → MCP Servers

Windsurf

Check Windsurf documentation for MCP configuration.

🎨 HTML Output

Generated HTML includes:

• Clean, modern design
• Tabbed navigation
• Markdown rendering (via markdown-it)
• Syntax highlighting for code blocks
• Responsive layout
• File metadata (last modified dates)

🔗 Links

• npm Package: https://www.npmjs.com/package/derek-memory-mcp
• GitHub Repository: https://github.com/dereknguyen269/derek-memory-mcp
• Issues: https://github.com/dereknguyen269/derek-memory-mcp/issues

🤝 Contributing

Contributions welcome! This is a simple, focused MCP server for sharing project specifications.

📄 License

MIT

Made for better AI-human collaboration 🤖❤️👨‍💻