Interesting idea, although it would be nice if Github optionally fixed spelling mistakes for you. Pull requests just for spelling errors have got to be annoying. (I just run flyspell in Emacs, so I never make spelling mistakes. Accept when I use the wrong word.)
The code is also pretty interesting. It's old-style Perl with indirect method calls, bareword filehandles, no "use strict", and so on... but then it uses modules that use Moose and MooseX::Traits. So we have a collection of scripts that look like they are from 1987, but that depend on Moose. Excellent!
(Actually, this inspired me to read the source code for Net::Twitter to see if it was using MooseX::Traits. Turns out they just copy-pasted from an older version of MX::Traits to get their traits functionality. Weird!)
So git becomes the presentation component of a spellchecker - a novel and interesting re-architecture.
You could do similar with peep-hole optimizations, bizarre bots that transform iteration to recursion (and vice versa), transforms to immutability, factoring-out common code, adding automatically constructed test cases ... I wonder what else.
It would be like helpful spam. EDIT or clippy "It looks like you're writing a loop. Would you like help?"
Actually, this could be a channel for code analysis as a service. You subscribe (for fee or free), it's all hosted and updated remotely and git-based (github public/private, or self-hosted).
The code is also pretty interesting. It's old-style Perl with indirect method calls, bareword filehandles, no "use strict", and so on... but then it uses modules that use Moose and MooseX::Traits. So we have a collection of scripts that look like they are from 1987, but that depend on Moose. Excellent!
(Actually, this inspired me to read the source code for Net::Twitter to see if it was using MooseX::Traits. Turns out they just copy-pasted from an older version of MX::Traits to get their traits functionality. Weird!)