i'm going to assume that i fixed it. approving my own fork.

i think that i have found the cause of the bug and fixed it in this fork, but i still dont know Haskell so i'd be happy if an exeperienced Haskell programmer could review it

i am unable to comprehend Haskell so I cannot fix the issue; if someone is able to fix this I will gladly accept a fork

Not very. It's so small because it has virtually no brains.

Major advantage is it's easy to prove correct, and ( within reason ) who cares about CPU time anyway?

Agree

To be fair, the type definition is Int -> Int, not (Num a) => a -> Int, so a Double would never be allowed into the function.

The Kata description is not clear.

I think there is no problem with the description except too many people do not read it carefully.

How is it not clear? How should it be improved?

I wonder if it is the same i am failing, i do not understand why it is not valid.

    +-+    
    | |    
    ++++   
    ++++   
   X+++    
     +---X 

Someone already reported similar problem for Haskell before: https://www.codewars.com/kata/5e28ae347036fa001a504bbe/discuss#5ea00083d751da0028221160 and it was not addressed, so you can raise an appropriate issue here. Just write a post again and mark it with "Issue" label. This will draw attention of other power users to the problem and hopefully someone would be able to fix it.

In what language?

can I see the corresponding mazes? I cannot find my mistake.

Just print (language specific) the parameter passed and look at the Kata output tab to see it.