I’ve documented a POST endpoint in my schema that requires a download link to be passed in as a parameter. I need chatGPT to generate this download link and send it over with the POST request. However, it frequently sends an empty parameters section or incorrect parameters like file_Id instead of download_link.
I’ve tried updating the description and context, using the openaiFileIDRefs array, and marking the parameters as required.
Has anyone else encountered this, and how did you ensure the GPT consistently called the endpoint with the correct parameters?