TokenTamer A proxy that reduces LLM token usage through context compression

Is there a way to bypass the tamer? i.e. if the implementation is actually needed for some subtle task