I don't have enough context to answer your question. Could you clarify what specific LLM harness or framework you're referring to? For example: - Are you talking about **LM Evaluation Harness** (by EleutherAI)? - **LangChain's `LLMChain`** or similar frameworks? - **OpenAI's API** or another specific platform? - A different LLM evaluation/harness library? Once you clarify, I can look up the current information and provide a detailed answer about middleware hooks.