

It’s convenience over security, something that creeps in anywhere there is popularity. For those who just want x or y to work without needing to spend their day in the terminal - they’re great.
You’d expect these kinds of script to be well tested against their targets and for the user to have/identify the correct target. Their sources should at least point out the security issue and advise to grab and inspect before straight up piping it though. Some I have seen do this.
Running them like this means you put 100% trust in the author, the source and your DNS. Not a big ask for some. Unthinkable for others.
On Dell server hardware with the right cards/licensing, you can remove the need for physical access to the server to input an FDE password by leaning on iDRAC. This provides access to the console remotely during the boot process (and thereafter).
Alternatives exist that supposedly do the same thing, but I’ve never had to try them. Airconsole, pikvm, blikvm etc.
You can keep this interface unexposed by using wireguard to dial in when you’re away, as per your original thinking. Just make sure the endpoint isn’t on the server you’re rebooting…