What problem does it solve? Customer interviews often fail because questions telegraph the hoped-for answer, producing polite agreement instead of real evidence. This Skill turns falsifiable hypotheses into open-ended, unbiased interview questions, each treated as a miniature experiment that tests a specific hypothesis without leading the witness. ## Core Features & Use Cases - Hypothesis-to-question crafting: Takes a single hypothesis or a full HYPOTHESES.md file and drafts questions that satisfy four criteria: confirms or negates the hypothesis, avoids hinting at an answer, elicits specific facts, and invites follow-up detail. - Grilling and iteration: Every draft question, whether written by the user or the AI, is attacked against the four criteria plus a polite-stranger test and a one-answer check before it is accepted. - Resumable file mode: Maintains a QUESTIONS.md file with numbered questions mapped to H-numbers, an in-progress header for resuming sessions, and a final assembly pass covering coverage, fit, order, and consistent customer vocabulary. - Use Case: A founder with a HYPOTHESES.md listing beliefs about how small-business owners handle missed calls uses this Skill to produce a one-hour interview script where each question tests specific hypotheses without biasing answers. ## Quick Start Ask the assistant to turn the hypotheses in your HYPOTHESES.md file into unbiased customer interview questions and save them to QUESTIONS.md.
