This prompt should be sent as part of a chain, with the input from the previous LLM response being fed into this reflective prompt.