You can open any finished chat, or fork any point of the discussion and simply ask:
Analysis of the rules that contributed positively or negatively to your decision or workflow in doing X to the Y.
It should list all levels of rules that may apply and how. This needs to be run fresh, and worded in a way that matches the Agent output. Without rules, everything is garbage, especially Python project structures. Luckily, for Rust, it has a specific respectful pattern, but in Python, any small change may cause it to spawn a new completely independent duplicated separate script file.