What problem does it solve? Writing implementation-ready specifications from a product requirements document is slow and error-prone, and ambiguous specs turn into bugs and misaligned expectations. This Skill decomposes a PRD into logical components, drafts a structured SDD spec for each one, and scores each spec's quality so gaps are fixed before coding begins. ## Core Features & Use Cases - PRD Decomposition: Reads _reversa_sdd/prd.md and proposes a component breakdown, confirmed with the user before drafting. - Structured Spec Generation: Writes specs from a 15-section RFC-style template with numbered requirements (RF-XX), non-goals, edge cases, and acceptance criteria, all stamped with the 🟡 planned seal. - Automated Quality Scoring: Runs scripts/spec_scorer.py (or a manual rubric fallback) to grade each spec 0-100 across completeness, testability, clarity, scope, and edge cases, iterating up to 3 times per component. - Use Case: After drafting a PRD for a new product, invoke this Skill to produce per-component specs in _reversa_sdd/sdd/, review the consolidated score report, then hand off to /reversa-forward for implementation. ## Quick Start Ask the agent to run /reversa-spec-sdd to decompose the PRD in _reversa_sdd/prd.md into component specs and score each one.
