Claude Code auto mode blocked 89% of dangerous commands in a controlled test. Humans blocked 13.6%

13.6% is the number that stings: a human clicking through approval prompts is a weaker gate than a classifier that never gets tired. Of the three, a Stop hook that runs the tests before the session may end is the cheapest to add. Did the test say how much of the missed 11% was actually destructive?

the 89% versus 13.6% gap makes permission boundaries worth testing rather than assuming