The article discusses the author's reservations about the increasing use of coding agents and loops in software development, citing concerns about code quality, maintainability, and the potential loss of human understanding and control. The author acknowledges the benefits of loops in certain domains, such as porting code and performance explorations, but expresses unease about their use in writing lasting code. The author also notes that opting out of this machine-driven future may not be an option, as security and competitive pressures may force developers to adopt these technologies. The article concludes by highlighting the need for better tooling and more nuanced approaches to harnessing the power of coding agents.