LLMs can learn if you provide it rules in your repo, and update those rules as you identify the common mistakes the LLM makes