Recommendation Engine

A solid starting point if you're building personalized recommendations from scratch. Covers the main approaches: collaborative filtering, content-based, matrix factorization, and hybrid methods. It's designed for typical use cases like e-commerce products, streaming content, or news feeds, and includes guidance on handling cold start problems when you don't have much user data yet. The skill also touches on evaluation metrics like precision@k and NDCG, which matter when you need to actually measure if your recommendations are working. It won't architect your entire system, but it gives you the prompting framework to get Claude implementing these algorithms instead of writing boilerplate recommendation logic yourself.