So where do the probabilities come from?
Training
Adjusting a model's numbers over a huge amount of text until it gets better at predicting what comes next. When training ends the numbers are frozen and shipped. Using the model doesn't change them.
Correcting an agent teaches the model nothing. Your correction is text in one conversation, and when that ends it is gone. For a lesson to reach the next session, it has to be written where the next session will read it.
Say it to your agent
“You won't remember this correction next session. Add it to AGENTS.md as a rule with one example from this repo, and show me the diff.”