"Let's think step-by-step" zero-shot CoT trigger — is paired with → Chain-of-Thought Prompts for RL-CAI (Appendix E.2)

lets-think-step-by-step-prompt alone only signals that reasoning should happen before an answer; it says nothing about what a harmlessness judgment’s reasoning should look like, what a constitutional principle is, or how a conversation and two labeled responses should be laid out. rl-cai-cot-prompts supplies exactly that missing scaffolding as hand-written few-shot examples, each with a substitutable principle placeholder, pre-written conversation, responses, and a worked chain of thought. The paper uses both together in a single prompt rather than either alone: "we use the ‘Let’s think step-by-step’ prompt... to elicit the chain-of-thought. In addition, we prepend several hand-written, few-shot examples in the same format" (constitutional-ai, §"Chain-of-Thought Prompting", p. 11). The trigger phrase starts the reasoning; the examples shape it into something usable.