This package monkey-patches the indentation logic in atom in an attempt to come up with something more, uhm, sane. The main difference betweenthe logic in this package and in the language-mode package (where the identation logic resides in atom) is that this package does a strict count of opening and closing scopes and uses the tokenizer to identify whether scope delimiters appear in a string or comment (in which case they are ignored -- which seems sane). Hence the resulting indentation will never get out of wack, meaning that the identation does not grow arbitrarily for longer files -- something that sometimes happens with the existing logic.
This package is transitional. Once we can come to an agreement on how to move forward with this in atom/atom this package will go away.