review

Multi-perspective code review loads persona-specific reference files—such as Fowler, Carmack, or React core reviewers—before evaluating code. Each persona applies its own priorities to identify issues, cite relevant lines, and deliver structured findings with severity and recommendations, so you get diverse viewpoints in one review.

AI-assisted review centralizes reviewer personas through reference files that encode each expert's principles. Load the appropriate references before review, apply persona guidance to flag issues with citations, and synthesize findings across perspectives to produce consistent, citable feedback aligned with your goals.

Yes. Citable feedback is core to this review approach—each finding references specific files and line numbers from your loaded persona references and the code under review, so recommendations are concrete and traceable.

Contextual evaluation maintains your stated goals when applying persona guidance. Before review, specify whether you're prioritizing bugs, architecture, or performance; the loaded persona references then steer critique toward those concerns while citing relevant principles.

Available personas include Martin Fowler (design patterns), John Carmack (performance), Anders Hejlsberg (language design), React core principles, AI visionaries, and others. Load persona reference files lazily on demand to activate the viewpoint you need for your specific review context.