“Anthropic's Claude code review feature utilizes over 10 parallel AI agents running for 20 minutes to identify bugs in a codebase.”