A cruel irony of coding agents is that everyone who blew off automated testing for the past 20 years is now telling the AI to do TDD all the time. But because LLMs were trained on decades of their shitty tests, the agents are also terrible at testing.
