Audit an LLM prompt template and traffic pattern for cacheability. Recommend prompt restructure, TTL choice, parallelization fix, and semantic-cache threshold.
