The broken windows theory of coding agents
A software engineering team's shift to optional code reviews after April, combined with AI coding agents, caused review counts to drop to almost zero and let "passable" one-off agent implementations become replicated gol…