I actually might’ve found a solution (shortly after last post). Part of the problem was how ChatGPT was sending the prompt
After re-reading the docs and seeing the 8k character limit for description_for_model I decided to go more in depth in my description for the model (in natural language). It seems to have fixed that issue, specifically describing this process:
A plugin that takes in whatever the user types in as a prompts as a json file, transforms it into text…
hope that helps!