When users submit queries to the Language Model (LLM) application, like customer support inquiries, the application must crank out a sequence of prompts just before presenting the question for the language model. The ultimate request on the language model usually includes a developer-hardcoded prompt template, crucial knowledge fetched from exterio