Very cool!

There’s a lot of room for improvement, though.

First of all, your code is largely flat. Granted, few things get used more than once, but breaking it up into a few well-named functions could make it much easier to read, regardless.

Also, rather than grope around in the POD anew for every question, I’d read all the data into a hash at the start of the script. That would make the different parts of your code much more focussed; possibly to the point of obviating the need to break it up.

Makeshifts last the longest.