Day 2 of the OpenAI sprint: negative overall.

2.1 is a decent feature. Auto-review for risky actions sounds useful, especially for sub-agents and long-running tasks. But I’m always on bypass permissions / YOLO mode anyway. My agent has never deleted anything or done anything bad, so I trust it completely. Not really useful for me.

2.2 is just qualification for higher API rate limits. Tier trash. I genuinely don’t understand why this is being announced as a feature. Feels relevant to a tiny number of enterprise users.

A reset would’ve been better. Homework rejected today.