many such cases. i was just working on getting a more accurate realized/estimate of slippage when the my execution engine is filling positions and CC assumed i was calculating it wrong.

i was pretty sure i didnt because i reconcile this data points manually quite often. i had to go into hyperliquid ide and get a bunch of execution prices to dump into the prompt for him to find that i was doing it right.
two things that come from this:
> never assume what CC says is right off the bat. verify the code, i know its boring, but in important tasks, do it.
> document these little quirks. after this is done, im going to dump this into the execution engine architecture md file, so it never forgets about it.
AI code gets a rep for slop and its true. you still have to be the person checking if the code makes sense and matches your desired output.