What problem does it solve? Vague or underspecified requests lead to misaligned implementations. This Skill turns broad ideas into explicit, testable specifications by interviewing the user one question at a time, scoring ambiguity quantitatively, and refusing to hand off to execution until clarity thresholds and readiness gates are met. ## Core Features & Use Cases - Socratic Interview Loop: Asks one targeted question per round, pressure-tests answers with evidence, assumption, and tradeoff follow-ups, and re-scores weighted ambiguity after each answer. - Depth Profiles & Gating: Supports --quick, --standard, and --deep profiles with distinct ambiguity thresholds and round caps, plus mandatory Non-goals and Decision Boundaries readiness gates. - Spec Crystallization & Handoff: Writes interview transcripts to .omx/interviews/ and execution-ready specs to .omx/specs/ for downstream planning and execution modes, with a specialized --autoresearch intake mode. - Use Case: A user says "improve the portfolio site performance." The Skill inspects repo docs, asks intent and scope questions across several rounds, then produces a spec with acceptance criteria, non-goals, and decision boundaries ready for a planning workflow. ## Quick Start Ask the AI to run a deep interview on your vague feature idea, for example by saying "deep interview me on adding a dark mode toggle to the site" and answer each question until the spec is crystallized.
