What problem does it solve? Open asks (pending questions awaiting human judgment) accumulate in the CALM memory database, and answering them ad hoc risks skipping items, exposing internal IDs, or letting the AI fabricate answers on the human's behalf. This Skill enforces a minimal, disciplined flow: list open asks once, present each to the human, and record only the human's actual answer. ## Core Features & Use Cases - One-shot open ask listing: Calls get_asks() exactly once with default filters, reports when no open asks exist, and notes when only the top N of M items are shown. - Human-driven answering: Presents each ask by title, question, and context (never internal IDs), then records the human's own words via answer_ask without AI-generated content. - Strict scope boundaries: Does not perform triage (triage_ask), withdrawal, meta-ask placement, or continuous monitoring, delegating those to ask-compose, ask-distill, ask-watch, and rule-placement. - Use Case: You return to a project and several judgment-pending asks have piled up. Invoke this Skill to walk through them one by one, answer each in your own words, and get a completion report noting that activity blocks remain until triage. ## Quick Start Ask the AI to list and answer the open asks, for example by saying "askに答える" or "open ask消化して".