reversa-spec-sdd

Spec-driven development decomposition works by reading a product requirements document, proposing logical components, and iterating until confirmed. It then generates individual SDD markdown files with explicit acceptance criteria, edge cases, and quality gap analysis.

Spec-driven development decomposition works by reading a product requirements document, proposing logical components, and iterating until confirmed. It then generates individual SDD markdown files with explicit acceptance criteria, edge cases, and quality gap analysis.

To generate SDD specs from a PRD, run the skill after your product requirements document exists at the required input path. Follow the component confirmation prompts to approve the decomposition, and the tool will draft, format, and score the specs automatically.

Yes, you can evaluate an existing spec or generate a single spec from provided input in standalone mode. This optional mode operates outside the main project pipeline, allowing you to independently score and analyze individual spec documents.

Python is required to run the spec quality scoring script, which evaluates specs on a 0–100 scale. If Python is not available, the skill will still generate the SDD markdown files but will skip the automated quality scoring and iteration loop.

SDD markdown files must include mandatory PLANEJADO markers for all spec items and contain a final evaluation report. This strict formatting ensures all requirements, edge cases, and quality gaps are explicitly tracked before handing off to the next development stage.