reversa-reviewer

To validate reverse-engineering specs, you review each generated unit for missing canonical files and internal consistency. This process detects cross-unit contradictions and reclassifies confidence levels across the full output set.

Cross-unit contradiction detection identifies conflicting assumptions and missing modules across the full generated reverse-engineering output set. It systematically challenges generated specs to reduce the risk of incorrect or incomplete specifications.

Generating human-validation questions involves applying confidence rules to inspect unit canonical files and optional artifacts. It outputs a structured questions file specifically targeting unresolved critical gaps found during the review phase.

Yes, applying confidence scoring reclassifies spec gaps by inspecting unit canonical files and applying confidence rules. It produces a confidence report and a structured questions list for any blocking gaps requiring human validation.

Reviewing per-unit spec folders requires reading state.json and config.toml to inspect unit canonical files. It checks business-rule coherence and generates human-validation questions with correct gap handling for the output folder.

A blocking gap confidence report is produced when applying confidence rules detects unresolved critical inconsistencies in the generated output. It surfaces structured human-validation questions to challenge incorrect or incomplete specifications.