What problem does it solve? Gathering requirements and aligning on user stories often suffers from unasked questions, assumptions made on the user's behalf, and premature conclusions. This Skill structures AI-human interviews so every open aspect of a topic or story is explored, asked about individually, and confirmed before work proceeds. ## Core Features & Use Cases - Explore-before-asking: Reads the knowledge base, PRD, architecture docs, and codebase first so already-answerable facts are never asked as questions. - Two modes: interview mode explores a freeform topic into a raw requirements blob; sync mode systematically covers six story aspects (goal, acceptance criteria, edge cases, dependencies, design, risks) until explicit shared understanding is confirmed. - Recommendation-driven questions: Every question ships with a proposed answer, and "don't know" replies become flagged assumptions rather than blockers. - Write-free synthesis: Returns the synthesis to the caller or conversation, optionally saving a handoff file in .pair/working/ on interruption. - Use Case: A product manager refining story #142 invokes sync mode; the Skill skips aspects already covered in the draft, asks about missing risks and dependencies one at a time, and returns an alignment synthesis mapped to the user story template. ## Quick Start Ask the AI to grill you on a topic, for example: grill me on the caching strategy requirements for the API layer.
