The End of Code Review: Coding Agents Supersede Human Inspection

A paper by Martin Monperrus argues that LLM-based coding agents have reached a capability threshold where they can replace traditional human code review as a quality gate in software development. The author contends that coding agents can fulfill all stated goals of code review more cost-effectively and with higher throughput than humans, and that having humans review AI-generated code is neither scalable nor provides meaningful assurance. This challenges a five-decade-old software engineering practice formalized in 1976.