What problem does it solve? Requests often name an outcome while leaving product decisions open: what counts as done, which data it stores, or which architecture carries it. Coding through those unstated decisions produces work nobody asked for. This skill turns an outcome into a decided brief by asking only the decisions the user would notice, deciding routine ones itself, and recording every answer with who closed it. ## Core Features & Use Cases - Decision map interview: Maps every open product decision, closes the ones the repository already answers with path evidence, and asks the rest in rounds of at most four questions with recommended options. - Two modes: Explore mode compares directions with stated trade-offs when the user asks for options; Specify mode grills, checkpoints, and writes the brief when the user wants the outcome built. - Brief storage: Writes the decided brief to docs/specs/<topic>.md or a GitHub issue marked with an exo:spec header, so a later session resumes from the stored copy after a context clear. - Use Case: A user asks to add a data export feature without saying what format, which records, or where it starts. The skill asks those three decisions in one round, settles file placement from repository conventions, and writes a brief with goal, decisions, acceptance checks, and proof seam. ## Quick Start Ask the AI to shape the outcome you want built, for example by invoking /exo:shaping with a short description of the result you have in mind.