discusslisted

# Discuss You reveal gray areas, make implicit assumptions explicit, and capture user preferences — BEFORE the concept is created. ## Execution The Main Agent performs `/discuss` **itself** — do NOT delegate it to a Sub-Agent. This Skill depends on direct interaction with the user via `ask_user`. A Sub-Agent cannot interact with the user. ## Why a separate Skill? `/analyze` opens up the problem space. Between the analysis and any external research, the user holds knowledge that no search can find: hidden assumptions, unstated preferences, and implicit decisions. `/discuss` surfaces those **first** — gray areas, unresolved questions, missing user preferences — so the subsequent `/research` round only asks questions the user genuinely cannot answer, and `/conceptualize` works on a solid foundation. ## Approach ### Phase 1: Identify gray areas Read the research results and the problem analysis from `/analyze`. For each topic, identify the **gray areas** — places where several sensible paths exist and the user needs to decide. **Gray-area types by topic:** | Topic area | Typical gray areas | |---|---| | Architecture | Where do new components live? Which abstraction level? Which coupling? | | API/interfaces | Response format, error handling, versioning, breaking changes | | UI/visual | Layout, density, interaction, mobile vs. desktop | | Workflow/process | Order, required vs. optional, automation vs. manual | | Data model | Schema design, normalization, migration, defaul