What problem does it solve? Turning a loose feature idea into a structured requirements document is slow and error-prone, especially when the new feature must align with an existing legacy system. This Skill converts a free-form sentence or paragraph into a complete requirements.md anchored in artifacts extracted from the legacy codebase by the reversa pipeline. ## Core Features & Use Cases - Context-grounded drafting: Reads architecture.md, domain.md, inventory.md, and code-analysis.md from the reverse-engineering output folder so every requirement cites real system sources. - In-progress feature detection: Inspects physical artifacts (requirements.md, roadmap.md, actions.md checkboxes) to detect an unfinished feature and asks the user whether to continue, pause, or abandon it before writing anything. - Guided quality loop: Applies a quality template checklist with up to three self-review iterations, limits open questions to three [DÚVIDA] markers, and suggests the next pipeline step (/reversa-clarify or /reversa-plan). - Use Case: You type "/reversa-requirements add OAuth login to the legacy portal" and receive a fully structured requirements document in _reversa_forward/001-oauth-login/ with citations to the legacy system's documented components. ## Quick Start Ask the agent to run /reversa-requirements followed by a sentence describing the feature you want to specify.
