Starting from the design phase of integrating MONAILabel into MITK, we have always tested MONAILabel endpoints at model level. Especially after finding discrepancies at model's data-metadata, we are on an conservative approach in supporting models.
The approach has led to the changes in D875 that only shows "supported models", almost only from Radiology app.

This approach is fine when dealing with only one app. But as the interest in different apps grows, how can we sustainably support and expose a lot of models? For eg. MONAILabel Bundle app contains 10 new models or a custom app with custom models can be created which won't be supported by MITK at all.
So how should we handle such situations?
Should we just switch to an approach where we actively block the apps that we are convinced not to be exposed but let the untested ones show in MITK on user risk?