classification

Classify incoming content before calling the LLM

The pattern is to classify and sort incoming content first, then call the LLM only when needed, reducing unnecessary model requests.