Hacker News new | ask | show | jobs
by nunez 23 days ago
AI shouldn't write tests. At least not all of them. Definitely not e2e's. The tests should be guardrails to constrain agents. This way, the author of code matters less.