What problem does it solve? Writing a PRD from scratch often means skipping research, making unexamined assumptions, and drifting into technical implementation details too early. This Skill enforces a structured brainstorming workflow that grounds every PRD in codebase context, market research, and explicit user decisions before any document is written. ## Core Features & Use Cases - Parallel Context Discovery: Runs codebase exploration and web market research simultaneously, then merges findings before asking any questions. - Structured Question Protocol: Asks one multiple-choice question at a time, focused strictly on WHAT users need, WHY it matters, and WHO the users are — never on implementation. - Approach Selection with ADRs: Presents 2-3 product approaches with trade-offs and records the chosen direction as an Architecture Decision Record in .compozy/tasks/<slug>/adrs/. - Canonical Template Output: Generates a complete PRD with mandatory sections (Overview, Goals, User Stories, Core Features, Phased Rollout, Success Metrics, Risks, ADRs, Open Questions) saved to .compozy/tasks/<slug>/_prd.md. - Use Case: A product manager wants to spec a new notification feature. The Skill researches the codebase and competitors, asks clarifying questions one at a time, presents scoped approaches, and produces an approved PRD ready to feed into a TechSpec. ## Quick Start Ask the AI to create a PRD for your feature idea, for example: create a PRD for a member attendance tracking feature.
