Description
Can you add a user_prompt_processor to the agent? I hope to be able to add DocumentURL to the user message list. However, the LLM model actually does not support processing documents. I can use the user_prompt_processor to convert the documents into text before sending them to the model , but when storing historical records, DocumentURL user_prompt are still stored, it is also possible to cache the user prompts that have been processed by the user_prompt_processor for use in message_history
References
No response