The difference is that the error messages contain instructions on how to resolve the issue.

Just flagging big functions will make the agent write small functions ~ but not necessarliy in a good way

(for example the agent might just cut `doOneThingAndTheOther` in half and call the second half `doOneThingAndTheOther2`)

[dead]