I admit I spent more time on this than I expected. It's a tiny bit frustrating to optimize prompts because the behavior of the LLM varies; a prompt might work o

I admit I spent more time on this than I expected. It's a tiny bit frustrating to optimize prompts because the behavior of the LLM varies; a prompt might work one time, but not the next. Still, seems like my prompting skills have some ways to go :)