What problem does it solve? Non-trivial feature requests often contain unresolved product, architecture, API, or persistence decisions. Implementing without resolving them forces the AI to guess, producing rework. This Skill enforces a disciplined workflow: inspect the repository first, classify uncertainties, and ask the user only about material decisions that the code cannot answer. ## Core Features & Use Cases - Repository-first discovery: Inspects the affected domain, contracts, tests, and existing patterns before asking any questions, treating the codebase as the source of truth. - Materiality classification: Sorts each uncertainty into repository-answerable, safe implementation detail, or material product/architecture decision, so only meaningful questions reach the user. - Structured interviews: Produces focused question rounds (at most 5-7 questions) with concrete options, recommendations, and trade-offs, plus a decision summary before coding begins. - Use Case: When asked to "add a new resource type with unclear lifecycle" to the Go CMS, the Skill inspects existing resource implementations, then asks targeted questions such as whether deleting a parent should cascade to child items, before any code is written. ## Quick Start Ask the AI to design and think through a new non-trivial Go CMS feature, and it will inspect the relevant code first and then present a short list of material decisions with recommended options.
