The correct option is Few-shot prompting using a handful of examples.

Placing three short snippets that illustrate the desired coding pattern inside the prompt is classic few-shot prompting. By giving the model several compact examples you guide it toward the same structure and style. This approach conditions the model on the provided demonstrations so it learns the convention you want it to follow in its next response.

Zero-shot prompting is incorrect because it involves giving an instruction without any examples. In this scenario the engineer is explicitly supplying examples.

Retrieval-augmented generation is incorrect because RAG augments the prompt with content retrieved from an external knowledge source. Here the examples are written directly in the prompt and there is no retrieval step.

Chat history utilization is incorrect because relying on prior turns of a conversation for context is different from placing curated example snippets in the current prompt. The question describes deliberate in-prompt examples rather than context from earlier messages.

Full AWS Practitioner Certification Question