code-review

Invoke the code-review skill against your current branch's changes. It audits the diff for structural regressions, abstraction quality, and spaghetti growth, then returns prioritized findings with concrete restructuring suggestions.

It checks for files crossing 1000 lines, ad-hoc conditionals bolted onto existing flows, feature logic leaking into shared modules, unnecessary wrappers and casts, and missed opportunities to delete complexity through restructuring.

Use it for deep quality audits of significant changes where long-term maintainability matters more than quick approval. Standard reviews catch bugs; this review catches design debt like tangled control flow and wrong-layer logic.

Files crossing the 1000-line threshold are treated as a strong code-quality smell because large files resist decomposition and become harder to reason about. The reviewer must either decompose the code first or provide a compelling structural justification.

It evaluates structure and design quality, not runtime correctness, security vulnerabilities, or test coverage. It also cannot verify whether a proposed restructuring truly preserves behavior, so suggested refactors still need test validation.