Some applications including older versions of Node will convert unicode characters into latin1 encoding by truncating the unicode values to fit into the 0-95 character space for basic latin characters. This script takes a text body in latin characters and returns unicode characters that when converted by a vulnerable application would translate back into the original basic latin text body.
This might be useful for performing http request smuggling and server side request forgery.
You can decode the payloads like this:
>>> "".join([chr(int(hex(ord(character))[-2:], 16)) for character in "ňɥɬѬů̠ʹɨѥɲͥġ"])
'Hello there!'