While a great feature, I never needed such a thing in Codex after GPT 5.2. It just one shots tasks without stopping
So we have proof by existence that this problem can be solved without any such mechanism. Wish to see the same relentlessness in Anthropic models
