What problem does it solve? Reverse-engineered specifications often contain contradictions, unverified inferences, and gaps that undermine reimplementation efforts. This Skill critically audits the specs produced by the reversa-writer, finds inconsistencies within and across units, reclassifies confidence levels, and generates targeted questions for human validation. ## Core Features & Use Cases - Spec Auditing: Checks each unit's requirements.md, design.md, and tasks.md for internal consistency, cross-unit contradictions, and missing behaviors. - Confidence Reclassification: Upgrades or downgrades 🟢/🟡/🔴 confidence markers based on code evidence, following defined classification rules. - Human Validation Workflow: Generates questions.md for unresolved gaps, processes user answers, and produces a final confidence-report.md with per-spec percentages. - Optional Cross-Review: When the Codex plugin is available, delegates an independent second-opinion review and merges its findings. - Use Case: After running a reverse-engineering analysis on a legacy codebase, invoke this Skill to validate the generated specs, resolve critical gaps with stakeholder input, and obtain a quantified confidence report before reimplementation. ## Quick Start Ask the agent to review the specifications generated in the reversa output folder and produce a confidence report with validation questions.
