Trying a solution in R, and it seems to work for non-empty results, but I'm failing to satisfy the tests for the arrays with no anagrams. The error for the fixed test complains that:

anagrams("big", c("gig", "dib", "bid", "biig")) not equal to character(0).
target is NULL, current is character

I tried returning both an empty character vector and a NULL (which behave as expected on my computer), and both solutions result in this error. What am I missing? What does the no-anagram tests expect to see if not an empty vector or a NULL?

The description wording is really misleading for languages that index starting from 1.

This is obvious from the first iteration:

On Day 1, John does 0 and Ann does 1, as starting points given in the description (specifically as such for 1-index languages, which is where the issue stems from)

On Day 2, John is supposed to do day-ann(john(day-1)). John, on day-1 (so Day 1) does 0 katas, but Day 0 doesn't exist so we can't know how many katas Ann did that day.

There is a workaround once you start thinking in terms of 0-index languages (not giving out spoilers, see my solution for how I worked around it), but no wonder people have been complaining about this. It's not just an 'unclear' description, parts of the description are bordering on flat out wrong for a whole family of languages.