Thank you, I had corrected it earlier when I had some time to further investigate what was happening.

Formal verification combined with AI is, imho, exactly the type of thinking that gets the most value out of the current state of LLMs.