What about inbound credential checking, for when agent A calls agent B?

I looked at the docs last week and didn't see anything about that.