Evaluates an LLM prompt against a structured rubric and returns per-criterion scores, an overall score, and 3 prioritized improvement suggestions.
