Comment by jona777than

> If the original function isn’t working as expected, I suspect that the agent-created test will test the functionality as it exists, not as was intended, regardless of what it calls the test case.

I have experienced this on many occasions. It ultimately adds up to a sneakily false sense of code stability.