clarify-questionlisted

# Clarify Question Use this skill when a request reaches Level 3 under `method/clarification-policy.md`, when the user explicitly asks to refine a question, or when a task must become a self-contained handoff. ## Role Your job is to discover the smallest set of user decisions that materially changes the work. Ask about goals and tradeoffs; investigate technical details yourself. Once the request is clear, produce a precise handoff prompt. Do not execute the original task while running this skill unless the user explicitly switches to execution. Do not turn clarification into authorization. ## Step 0: Load The Task Profile Follow `method/profile-context.md` before deciding what the user may need help clarifying: 1. Read the current conversation. 2. Load the configured base profile and only relevant Markdown memory. 3. Separate confirmed preferences from agent inference. 4. Record source, freshness, and confidence for conclusions that affect the questions. 5. If sources conflict, follow the documented precedence and expose the conflict. No profile entry is evidence that the user is unfamiliar with a domain. Upgrade only when there is a concrete hidden choice that could change the result. ## Four-Level Boundary - Level 1: execute directly. Do not invoke this skill. - Level 2: state one key reversible assumption, then continue without a full clarification round. - Level 3: explain why a decision matters, then ask 1-3 concise questions. - Level 4: stop the side effect an