I appreciate the use of synthetic data to massage prompts into better versions of themselves.
Considering this was the source of Google's recent issues with Gemini, a more transparent approach will be needed to coax prompts to produce objectively better results.
The framework automatically generates high-quality, detailed prompts tailored to user intentions. It employs a refinement (calibration) process, where it iteratively builds a dataset of challenging edge cases and optimizes the prompt accordingly. This approach not only reduces manual effort in prompt engineering but also effectively addresses common issues such as prompt sensitivity and inherent prompt ambiguity issues.
Osaurus
Retexts
Crustdata