@Marvin can you write a broad set of rules for the AI, like part of a prompt template, that I can inject into each prompt so have the LLM write code nicely in line with Prefect general guidelines for Python? I like using functional programming paradigms, no classes (limiting or eliminating side effects?), immutability, pure functions, but not lambdas everywhere since that is hard to read. Also I plan on deploying to GCP (either cloud Run or Vertex AI). My pipeline may or may not have some Langgraph agents called within subflows to perform tasks that only AI can perform.