The correct option is To express intent clearly and supply relevant context so the model can shape a helpful response.

To update Copilot with your company's code so it learns new patterns is incorrect because Copilot Chat does not update or retrain its underlying model from your prompts or your private repository during an IDE session. It can use available context to answer your question, yet it does not learn new patterns from your individual usage in real time.

Vertex AI is incorrect because it is a Google service and it is not related to the objective of prompt design for GitHub Copilot Chat.

To exploit temporary token context so Copilot returns exact code snippets is incorrect because prompt design is not about forcing verbatim outputs. The aim is to guide the model with clear goals and relevant context so it can generate helpful and trustworthy responses.

To lower response latency by enabling local caching in the editor is incorrect because prompt wording does not materially affect latency. Response time depends on service performance, network conditions, and IDE capabilities rather than how the prompt is phrased.

Full AWS Practitioner Certification Question