Run focused, repeatable code reviews using configurable presets and multiple LLM models in parallel. Findings become trackable feedback items with a verify-apply-resolve lifecycle.