[avocadoking]

Do you have any insight if LLMs sometimes get confused by your filters?

[tim-projects]

He says he adds an output message, but I've tried this myself and I find that quite a lot of the time the agent prefers its own internal monologue over the output of a command.