I'm confused by this also. That is, when I create a low power end node without alternate means of communication, it seems impossible to see its IP address. Maybe the border router maintains a list of granted IP addresses? All of the NXP examples seem to rely on typing in "ifconfig" into the console to learn IP addresses.
From what I read about CoAP, I might need to create a CoAP Resource Directory? My understanding is that end-nodes can POST their available resource links somehow into a /.well-known/core directory, and then outside sources can see those resource links, and GET them. If this is true, does the resource directory reside on the border router, or does each end-node have its own directory?
I'm using the Copper extension for Firefox to perform coap interaction, and I have an RNDIS border router set up for the connection.