Auto Mode will be the default in Claude Code – because humans can't be trusted (thenewstack.io)

[flagged]

I still don't see the appeal of Auto Mode. If I don't trust the model to run arbitrary commands, how much additional trust can be gained by an additional non-deterministic system. The point is that there are some failure modes which are completely unacceptable. Auto Mode lowers that risk but doesn't remove it. An unacceptable failure is still possible.

Auto mode works really well with Safehouse [1]. I usually call Claude like this:

    safe() {
        safehouse --env --add-dirs="/tmp:~/.claude" "$@"
    }

    safeclaude() {
        safe claude --enable-auto-mode --permission-mode auto
    }

This way I will never fear Claude `rm -rf`-ing my home directory.

[1] https://agent-safehouse.dev/

[flagged]

I would generally disagree with this unless it can be fine-user tuned (for the current session person allowance). Although in general; I believe this is letting ai take the wheel, which I 100% disagree with as more damage will likely be done without human oversight/retro-actions and even if caught would it be a humans fault?