What problem does it solve? Generic recommendation lists ignore individual taste. This Skill combines what the user has already said they like and dislike with current web search results to produce recommendations tailored to that specific person. ## Core Features & Use Cases - Taste Recall: Queries memory for prior likes and dislikes in a category before searching, treating dislikes as strong filtering signals. - Current Candidate Discovery: Uses web search and page fetching to recommend from options that exist now rather than stale training data. - Ranked Picks with Reasoning: Delivers 3-5 picks ranked by predicted fit, each with a reason tied to the user's stated taste, an honest caveat, and one labeled stretch pick. - Outcome Recording: Stores the user's reaction to each recommendation so future suggestions improve and rejected picks are never repeated. - Use Case: A user asks what to watch next. The Skill recalls that they loved slow-burn sci-fi but abandoned two action franchises, searches for current releases, and returns five ranked picks with reasons and caveats. ## Quick Start Ask the assistant to recommend something to watch, read, play, or try based on what it already knows about your taste.
