Hacker News Reader

I wonder if someone has tried to chain ChatGPT's output with static analyzers/compilers/linters in a loop to improve the final response.

Structural Alignment: Modifying Transformers (like GPT) to Follow a JSON Schema