What problem does it solve? Plans built on assumptions get rejected at approval gates and waste research passes. This Skill extracts the owner's actual intent through a structured one-question-at-a-time interview before any research, planning, or building begins, so downstream agents work from confirmed decisions instead of guesses. ## Core Features & Use Cases - Guided interrogation method: Asks one question at a time with a recommended answer, walking the decision tree so high-impact forks are resolved first. - Scoped question branches: Covers outcome and scope, users and value, budget/timeline/runtime constraints, security gate profile, data and integrations, design voice, and explicit out-of-scope items. - Discovery.md artifact: Writes answers, stated assumptions, plan implications, and open questions to potential-projects/<name>/Discovery.md, which research, viability, plan, and build agents then read. - Non-interactive fallback: When the owner cannot answer, marks every needed answer as ASSUMED — unconfirmed and surfaces them in the YOU DO list instead of stalling. - Use Case: At the start of a /kickoff for a new product idea, run this Skill to grill the owner on budget, target users, and must-not-change constraints, producing a Discovery.md that the research and viability agents consume. ## Quick Start Before planning my new project idea, interview me one question at a time about scope, users, and constraints, then write the results to Discovery.md.
