Interview Scenarios - System Prompt Extraction via Hypothetical Framing Follow-Up Questions

Source document: 09-interview-scenarios.md Reference scenario: 01-prompt-injection-defense.md -> Scenario 3: System Prompt Extraction via Hypothetical Framing

Scenario lens: Non-literal extraction attempts that use hypothetical, creative, translated, or reframed prompts to expose confidential rules. Document lens: interview preparation, persona-aware answers, and depth progression.

Use these prompts to push past the base scenario and explore deeper design, operational, interview, or storytelling tradeoffs.

Answer document: ANSWERS.md

Easy

1. If asked about hypothetical prompt extraction, what is the shortest answer that still makes clear why regex-only defenses fail?
2. Which concrete refusal pattern or compartmentalization decision would you name first to show practical depth?

Medium

1. How would your answer differ if the interviewer cares more about privacy leakage than about model jailbreaks?
2. What metric or test suite result would you cite to prove the mitigation worked beyond anecdotal red-team wins?

Hard

1. How would you explain the tradeoff between keeping system behavior configurable and minimizing the damage of partial prompt leakage?
2. What mistake would weaken your credibility if the interviewer asks whether secrets should ever live in prompts at all?

Very Hard

1. How would you handle a skeptical interviewer who argues that no extraction defense is trustworthy unless you prove non-reconstructability?
2. If the interviewer asks for a second example on the spot, how would you pivot to a related document without sounding like you memorized isolated stories?