The agent identifies specific taste patterns from the user's high-rated items and suggests new media that aligns with those preferences, grounded in the user's actual library history.
The agent first connects to the Achriom MCP server via OAuth. It then calls get_by_rating to retrieve items rated five stars. It analyzes these results to identify common themes, genres, or moods. Based on this analysis, it uses search_library to find new items that match the identified patterns. The agent ensures all recommendations are grounded in the tool results and does not invent items the user does not own or like.
- Category: Research and knowledge
- Effect: Read-only
- MCP servers: Achriom MCP (https://koskamo.com/explore/achriom-mcp)
- Source: Achriom MCP: Connect Claude, ChatGPT & Any AI (https://www.achriom.com/mcp)
- Source quote: ""Find the pattern in everything I've rated five stars. What does my taste actually want?""
- Playbook page: https://koskamo.com/explore/use-cases/analyze-taste-patterns-and-recommend-media
Prompt: Connect to my Achriom library. Find the pattern in everything I've rated five stars. What does my taste actually want? Then suggest one new item that fits this pattern.