ai-code-review-refactor

To verify AI-generated code APIs, check that imports, method calls, and parameter signatures exist in declared versions to stop runtime failures. This prevents hallucinated APIs from reaching production environments.

A safe refactor boundary ensures behavioral equivalence and surfaces silent changes to observable behavior as reviewable, separated work. This prevents unsafe behavioral changes during AI-assisted code modifications.

Review AI code for missing tests and type safety failures by demanding an evidence inventory and explicit test adequacy expectations. This grounds completion claims in validation outputs rather than plausibility.

AI code refactoring requires dependency and security auditing to harden reliability. This flags swallowed exceptions, silent null returns, insecure data access paths, and missing adversarial review where needed.

Prevent architecture drift and dependency pollution by reviewing AI-assisted changes for hidden assumptions and over-abstraction. Apply strict API verification and bounded secure refactor scope before accepting changes.

You need an evidence-based completion report when validating or approving AI-assisted changes across existing codebases. It ensures completion claims are grounded in validation outputs rather than plausibility, especially for behavioral changes.