Haha, it queries chatGPT. It required a good amount of prompt engineering to get reasonable results, and there's definitely plenty of room for improvement. Glad you liked it!
How does the play animation work? Do you get some sort of json blob output? How do you make sure it fits your schema?
I get it in the form of a Python array that I parse. I try to prompt-engineer to make sure that it is returned in the expected form, but you could probably try to break it with specific inputs.