Anthropic launched a Claude Code code review feature where, when a Pull Request is opened, multiple AI agents analyze the code in parallel, rank defects by severity, and provide feedback in the form of summaries and inline comments.