Against all hope, maybe someone from openai will see this and make needed fixes.
For some reason I cannot build a gpt that uses our plugin’s OpenAPI spec. The gpt attempts to invoke the plugin!
I have a copy gpt that accesses our development system ok. So I know the gpt and this OpenAPI spec can work.
The problem seems to be when using the same server url as our plugin. When I change the server url to another value, it at least doesn’t attempt to invoke our plugin.
We need this to work as our migration path for customers currently using the plugin so we can transition them to a gpt…