Not really the same issue.
For yours, you just have to consider it goes back to 6.
Even negative indexes could be defined that way.

Additionally, the return value is unspecified if all of the letters are shared between the two names (such as LIAM/MILA, CLAY/LACY, ROAN/NORA, etc.).

Based on the reference solution, "Siblings" should be returned, but the description uses 1-based indexing, meaning that an index of 0 should be undefined.

It loops back to 1

Description has to state explicitly what to do with names with multiple occurences of common letters.
Take the following example:

show_relationship('KEVIN', 'ANN') == "Enemies"
KEVIN and ANN have character `N` in common.
You must cross out all Ns in both names: 1 in KEVI~N~ and 2 in A~NN~.
After this step, only 5 letters remain.

Approved some time ago

One of the easiest to understand imo

I think there's a problem with the randomized test. Cause it says on the log:
HSREBQTK
FLLCG
13
1
We get the value 1 which obviously corresponds to "Friendship" but an error occurs and says:
expected:<[Siblings]> but was:<[Friendship]>
although on previous randomized test with the same result:
KRK
BOOZJDTZVI
13
1
There is no error
Please fix this.

Indeed, and in fact what is the expected outcome of that?

Since the F is counted as the first step, what should we consider "step 0"?

random tests added

What happens if the total remaining characters > 6 ?

Theres more too it than that bud. Follow the logic. You're saying everything that's not 0 and divisable by 3 is Java.

What about Coffe, CoffeeScript, and Javascript?

if (n != 0) {
      if (n%3==0 && n%4 != 0) {
        if(n%2==0)  {
          return "JavaScript";
        }
        else {
          return "Java";
        }
      
      }  // not devis 4
      
      else if (n%3==0 && n%4==0) {
        if(n%2==0) {
          return "CoffeeScript";
        }
        else {
          return "Coffee";
        }
        
      } // devise form  
          
    } // MAIN IF
    
    
    return "mocha_missing"; // ELSE!!!!!!!!!!!!!!!!!!

You really need randomized tests to make all these hard-coded implementations to fail.

Two test cases don't seem to evaluate properly.