Here you can find the repo if you want to use it. The jist of it is that you configure the source code like you would dwm, or st and then you have a very fast and light fetch program. For the ascii it takes a file called conf found in $HOME/.config/cfetch/ so you can easily change the ascii output. The repo has two ascii files one for gentoo and one for openbsd by default, check it out, critisize it do whatever.
Yesn't, yes it likey runs the same commands as neofetch, but if you run time with it you can see the difference, neofetch on my linux machine takes 0.3 seconds to run but cfetch takes >0.01 seconds to run, sometimes it doesn't show. In practice you probably wont notice it, but MINIMALISM
Not op, but functionally no, as it would be rare a user wouldn't use bash to run it but it does remove the dependance on a shell, and is likely quicker. It could, technically, be run without a shell too. It would also work on non posix compliant shells, or anything which can run the command. Not something that this application would be likely to do, it is just a fetch app, but it's still a fun little project