The refactor instinct is escape, not strategy. Spend 45 minutes with the actual bug: add logging, bisect the change history (`git bisect`), reproduce in isolation. 9 times out of 10 the bug is a 3-line fix once you understand it. Refactoring while you don't understand the bug just moves it somewhere harder to find.
Prompt to paste
Don't refactor. I have a specific bug: [describe it]. Help me debug it in place: suggest log points to add, a way to reproduce it in isolation, and which recent change to bisect first. Find the root cause before proposing any code change, and keep the fix as small as possible.