I wanted to know it was only returning source. My suspicions always go up when I have the LLM lean on its "deep memories". too much fluff, inconsistent translations, stuff like that.