That's a quite interesting concept, my way to do 'katas' is with projecteuler.net, now I'm getting some coding speed by reading a book about patterns or checking the wikipedia for software patterns. If you get to combine both, then sky is the limit!
PE has always struck me as more focused on math than programming. The easy problems handle brute force but the hard ones are generally bested by a mathematical understanding of the problem rather than programming prowess & algorithmic kung-fu. I've been meaning to check out Rosalind [1] to see if it opens up the door for more challenging programming or just shifts the problem domain to biology.
If project euler seems too "mathy" try some problems from Top Coder or similar programming challenges. I find they're pretty fun, and can be a bit of a puzzle to work out what's the most efficient approach.
I'm not saying that it's "too "mathy"", just that it's commonly trotted out as a good resource to develop programming skills when, IMHO, it's better suited to honing mathematical reasoning. All the single page programs in the world aren't really going to help somebody become better at writing real programs.
