prdlisted

# PRD — Carson Pattern, Strict The biggest mistake Carson identified: _"They try to rush through the context where you just don't have the patience to tell the AI what it actually needs to know to solve your problem."_ This skill prevents that rush. It refuses to write a PRD until the user answers clarifying questions, refuses to generate subtasks until parent tasks are approved, and refuses acceptance criteria that can't be machine-verified. ## When to use this skill **Mandatory** for: any feature estimated >1 day; anything touching billing, auth, data migration; anything user-facing that ships. **Skip** for: typo fixes, dependency bumps, doc edits, refactors with no behavioral change. `/bs:dev --next` and `/bs:dev <name>` auto-route to this skill when complexity assessment scores ≥ M (medium) or higher. Override with `/bs:dev --no-prd`. ## Usage `/bs:prd <feature-name>` — start a new PRD interactively `/bs:prd from <voice-recording-or-rough-notes>` — turn "verbal diarrhea" into structured PRD (Carson's preferred entry point) `/bs:prd review <path-to-prd>` — re-run validation on existing PRD `/bs:prd tasks <path-to-prd>` — generate dot-notation task list from existing PRD ## Phase 1 — Clarifying Questions (REQUIRED) Ask exactly **3 to 5 questions**, targeting the **most critical gaps** for understanding. Not 10. Not 1. Tailor to what's actually missing from the input. **Rule**: If the user's brief already covers one of these, skip it. Ask only what's actually gap.