Your question is Designing Hallucination-Resistant Prompts. Take a moment with it on the right.
Talk me through your thinking if you like. When you're confident, submit your answer and I'll grade it like a real screen (7/10 or better passes).
Explain your methodology for designing and testing system prompts to prevent model hallucination.
Discuss how you define the task contract, constrain unsupported claims, specify uncertainty and refusal behavior, and separate instructions from untrusted user or retrieved content. Explain how you would build offline and online evaluations, use adversarial cases, and balance hallucination reduction against helpfulness, latency, and cost. Include a practical example of a prompt and the tests you would run.