ai-recommendations
AI-powered recommendation engine specialist using Google AI SDK for personalized content suggestions
When & Why to Use This Skill
This Claude skill serves as a specialized AI recommendation engine assistant, leveraging the Google AI SDK to help developers build and integrate personalized content suggestion systems. It focuses on transforming raw user preferences into structured, high-quality recommendations while providing robust implementation patterns for error handling, smart filtering, and performance optimization.
Use Cases
- Personalized Content Discovery: Building a 'Recommended for You' feature for streaming or blog platforms by analyzing user history, favorite creators, and genre preferences.
- E-commerce Personalization: Implementing a product suggestion engine that intelligently filters out disliked items and ensures only available inventory is recommended.
- Developer Integration: Streamlining the implementation of Google AI SDK in TypeScript projects with pre-defined patterns for structured prompts, caching, and asynchronous request handling.
| name | ai-recommendations |
|---|---|
| description | AI-powered recommendation engine specialist using Google AI SDK for personalized content suggestions |
AI Recommendations Specialist
Instructions
When working with AI recommendations:
Generating Recommendations
- Collect user preferences from database
- Build context with liked/disliked content
- Include favorite actors and directors
- Use structured prompts for consistent output
AI Integration
- Configure Google AI SDK in
src/lib/ai/recommendations.ts - Use structured output for reliable parsing
- Handle API errors and rate limits
- Implement fallback recommendations
- Configure Google AI SDK in
Smart Filtering
- Exclude already recommended content
- Filter out user's disliked items
- Ensure content is available
- Apply content rating restrictions
Performance
- Cache recommendations to reduce API calls
- Batch multiple user requests
- Generate recommendations asynchronously
- Store results for quick retrieval
Examples
Generating recommendations:
import { generateRecommendations } from '@/lib/ai/recommendations'
const recommendations = await generateRecommendations({
userId: 'user-123',
preferences: userPrefs,
count: 10
})
Building AI prompt context:
const context = {
likedContent: user.preferences.liked,
dislikedContent: user.preferences.disliked,
favoritePeople: user.preferences.people,
recentlyRecommended: user.recommendations,
genres: user.favoriteGenres
}
Handling AI response:
const response = await model.generateContent(prompt)
const { text } = response.response
const recommendations = JSON.parse(text)
// Validate and filter recommendations