There are no random tests.

Only 3 common passwords (out of all 50) are tested in the fixed tests.

This kata is a typical "grab bag" kata: it does a million things of very varying difficulty that has nothing to do with each other.

Also, there are Elixir exclusive built-ins (e.g Jaro distance) involved, so it'll have issues translated to other languages.

This is very bad password validation. It definitely shouldn't be used anywhere in production.

This is not a question

Brilliant!

The best solution would probably be to permanently associate a solution with a language version but I guess they didn't go woth that.

Right, the comment was meant for someone who might be confused as to where the function went :)

I can't speak for others, but if I recall correctly I submitted this solution around 2016 or 2017 when codewars was using an ancient version of elixir (something like 1.2). Obviously you would not use String.strip with 1.9 because it doesn't exist.

String.strip was deprecated in favor of String.trim, String.trim_leading, String.trim_trailing

Best practice.

did the same exact thing)))

nice

yeah, it doesnt filter the entire array twice

The random tests don't check for false positives. Returning words is enough to pass.
Your solution - LOL. But it's incorrect: the simplest counterexample is "dy" and "nn".
There's Elixir 1.6 now, so to_char_list should be changed to to_charlist.
And I think the example in the description should be more similar to those in other languages.