It's necessary to allow Claude Code to be fully autonomous, otherwise it will stop and ask you to run commands.

and just letting it to do whatever it thinks it should do, without a human intervening, is a good plan?

Depending on the breadth (and value) of the sandbox: Sure? Why not?

To extend what may seem like a [prima facie] insane, stupid, or foolhardy idea: Why not send the output of /dev/urandom into /bin/bash? Or even /proc/mem? It probably won't do anything particularly interesting. It will probably just break things and burn power.

And so? It's just a computer; its scope is limited.

Discovering that is the entire intent of this experiment, yes.

fair point. will re-read the whole thing. I'm sorry for my ignorance.

the "best practice" suggestion would be to do this in a sandboxed container