Unfortunately, it doesn't look like the code supports having multiple backreferences in a row... Logically, something like this should work:

which should generate the 4 characters at the front... then copy them to the end in reverse order... for example:

Unfortunately, it seems that the tokens parser is prioritising the ";filename;" parsing... and treating ;9b%; as a filename for a keyboard walking map pattern and spits out an IOError


I'm not sure of a workaround for this, that doesn't involve a massive pre-generated token file that has the 4 characters matched with their reverse 4 characters etc.