wht: refining LLM response by coerced reframing.

why: to fine-tune with unstructured context.