Evaluate and improve any AI prompt (prompt_a) through a staged, evidence-based pipeline. Functional evaluation checks whether the prompt follows rules, output contracts, quality requirements, and safety boundaries. Optional effect evaluation checks whether outputs work for intended readers through