Just brain storming... could another possible solution be to create a public list of hashes of well known scripts, and then write a small `sh` wrapper that verifies that hash of the script it's about to execute?
You shouldn't always trust even 'trusted' scripts due to the potential for various types of man in the middle attacks. True, though, that with all things being equal some scripts are more trustworthy than others, but if you're on a compromised network then you can't trust any.