This document describes the OpenCloset functionality that has been added to the Val Town ChatGPT App SDK Starter.

OpenCloset is a digital wardrobe management system that allows users to:

1. Upload photos of their clothing items
2. Automatically categorize them as tops or bottoms
3. Get outfit suggestions from their saved items
4. Browse their digital closet

• ItemCard - Displays individual clothing items
• MultiItemCard - Shows multiple items in a grid layout
• OutfitCard - Displays complete outfits with top and bottom pieces

• Cloudinary Integration - Handles image upload and storage
• In-Memory Store - Stores clothing items (MVP implementation)
• MCP Tools - Provides ChatGPT integration

Add these to your Val Town environment:

CLOUDINARY_CLOUD_NAME=your_cloud_name
CLOUDINARY_API_KEY=your_api_key
CLOUDINARY_API_SECRET=your_api_secret
CLOUDINARY_UPLOAD_PRESET=your_preset (optional)

1. Create a free Cloudinary account at https://cloudinary.com
2. Get your cloud name, API key, and API secret from the dashboard
3. (Optional) Create an unsigned upload preset for easier integration

Uploads and categorizes clothing photos.

Input:

{
  photoUrls: string[],  // Array of image URLs from ChatGPT attachments
  category: "top" | "bottom"  // Clothing category
}

Output:

• Single item: Displays ItemCard widget
• Multiple items: Displays MultiItemCard widget

Suggests an outfit from saved items.

Input: None

Output:

• Displays OutfitCard widget with randomly selected top and bottom
• Error message if insufficient items in closet

Lists all items in the closet.

Input:

{
  category?: "top" | "bottom"  // Optional category filter
}

Output:

• Displays MultiItemCard widget with filtered items
• Empty state message if no items found

{
  id: string,           // UUID
  imageUrl: string,     // Cloudinary URL
  category: "top" | "bottom",
  createdAt: number     // Unix timestamp
}

• CaptureItemsOutput - Single item or array of items
• SuggestOutfitOutput - Top and bottom item pair

1. Upload photos in ChatGPT
2. Say: "Add these to my closet as tops"
3. System uploads to Cloudinary and saves items
4. Displays ItemCard or MultiItemCard widget

1. Say: "Suggest an outfit from my closet"
2. System randomly selects one top and one bottom
3. Displays OutfitCard widget with the combination

1. Say: "Show me all my closet items"
2. System displays all saved items in MultiItemCard widget
3. Can filter by category: "Show me all my tops"

1. ChatGPT provides image URLs from user attachments
2. uploadToCloudinary() function handles the upload
3. Supports both signed and unsigned upload methods
4. Returns secure Cloudinary URL for storage

• Each tool specifies openai/outputTemplate in metadata
• Widgets automatically render based on structured content
• Consistent styling across all clothing-related widgets

• Cloudinary upload failures are caught and reported
• Empty closet states are handled gracefully
• Missing environment variables cause clear error messages

• In-memory storage using CLOSET array
• Data persists only during server runtime
• Fast and simple for testing and development

• Database persistence using existing SQLite setup
• User-specific closets with authentication
• Advanced categorization (colors, seasons, brands)
• Outfit history and favorites

1. Set up Cloudinary environment variables
2. Upload clothing photos in ChatGPT
3. Test each MCP tool with various inputs
4. Verify widget rendering and data flow

• "Add these photos to my closet as tops"
• "Add this to my closet as a bottom"
• "Suggest an outfit"
• "Show me my closet"
• "Show me all my tops"

1. Cloudinary upload fails: Check environment variables
2. No outfit suggestions: Ensure both tops and bottoms are in closet
3. Widget not displaying: Verify openai/outputTemplate metadata

• Check server logs for Cloudinary errors
• Verify image URLs are accessible
• Test Cloudinary credentials separately

• Database persistence
• User authentication and multi-user support
• Advanced item metadata (color, brand, season)
• Smart outfit suggestions based on weather/occasion

• AI-powered style recommendations
• Integration with shopping platforms
• Social sharing of outfits
• Wardrobe analytics and insights