The DSPy framework (https://github.com/stanfordnlp/dspy ; https://dspy.ai/) supports advanced prompt engineering and refinement, like iterative improvements to prompts based on RAG-based responses, and even tuning model weights to get better results for a given set of prompts. There is likely some overlap with CurateGPT functionality and it would be informative to compare the two. Building in DSPy support as a complement to CurateGPT agents could also be very configurable.