Cursor Setup - Deeper Dive

[rpattcorner]

It would be very helpful to know a little more about how shimmy "just works" with cursor.

For example, in the Cursor IDE, we can "add model". Doing so simply creates a "shimmy" entry in the models list that can be selected, but there is no additional field for the endpoint (http://127.0.0.1:11435/V1, etc). Or should a model that shimmy is intermediating be added as the model name instead? Neither seems to allow further configuration to the endpoint.

So ... guessing ... what's probably intended is to use "Add model" in conjunction with the widget to "Override the OpenAI Base URL"?

If so, that can be done, but immediately runs into the well known problem of cursor blocking private networks:

When I read "just works" with cursor, I figured shimmy had found some workaround for the necessity to add an ngrok additional shim to work around the private network block. Is this not so?