What problem does it solve? Writing requirements that are ambiguous, untestable, or disconnected from legacy source code leads to rework and lost traceability. This Skill reviews requirement statements against the EARS (Easy Approach to Requirements Syntax) notation so each one follows a single, testable pattern with a verifiable link back to its origin. ## Core Features & Use Cases - EARS Pattern Classification: Classifies each requirement into exactly one of six patterns (ubiquitous, event-driven, state-driven, optional, unwanted behavior, complex) using the standard "shall" (DEVE) templates. - Quality Checklist Enforcement: Detects common defects such as ambiguity, compound requirements, passive voice, hidden implementation details, and missing acceptance criteria. - Legacy Traceability: Requires every requirement to carry a REQ-ID and a non-empty source_legacy pointer to Natural programs (.NSN), Adabas DDMs (.ddm), or a justified [GREENFIELD] token, validated by the legacy-traceability CI task. - Use Case: During a legacy modernization sprint, paste a draft requirement like "the system must be fast" and receive a corrected EARS statement such as "QUANDO uma pessoa enviar um formulário, o sistema DEVE responder em até 500 ms" with a REQ-ID, acceptance criteria, and a source_legacy line pointing to the original Natural program. ## Quick Start Ask the assistant to review your draft requirements for EARS compliance and classify each one by pattern with a REQ-ID and legacy source trace.
