skill-staged-review

Staged code review is a validation process that separates spec compliance checks from code quality assessments. By applying sequential review gates, it ensures intended outcomes match implementation boundaries before evaluating architecture, error handling, and security in pull requests.

You verify spec compliance by applying a Stage 1 review gate that loads intent contracts and checks boundaries before proceeding. This ensures the PR fulfills intended outcomes prior to executing Stage 2 code quality assessments like architecture alignment and security analysis.

You need multi-LLM synthesis for code review when preparing major feature releases or high-risk changes that require rigorous cross-model validation. It delivers a unified quality verdict by aggregating multiple model perspectives on architecture, error handling, and security considerations.

Yes, staged review gates apply directly to high-risk feature releases and major PRs. The two-stage process validates intent contracts, detects stubs, and performs boundary checks, making it suitable for rigorous validation across software projects before merging.

The best way to detect stubs and check boundaries is applying a Stage 1 review gate that validates spec compliance against intent contracts. This automated boundary check identifies incomplete implementations before advancing to Stage 2 code quality assessment.