Subscribing agents to other agents


I was wondering if there’s some scheme on design for allowing agents to “subscribe” to other agents. As if an agent might or should have a form of “contact list”, some kind of simple “social” protocol where two agents can acknowledge they know each other, thus opening relevant channels for data transfer and maybe other capabilities.

Is this form of inter-agent interaction presently being considered? Are there any formal protocols being worked out for secure inter-agent “acquaintance” and communication?


Hello Carlos,

In XDI, we have done a lot of work on the kinds of interactions you are describing. We have “connection requests” (I ask something from you) and “connection invitations” (I offer something to you) to establish “link contracts” between entities in various ways. These requests can be manually approved/rejected, or automatically approved/rejected based on policies. Link contracts can have “push” semantics, i.e. if data covered by a link contract changes, it is pushed to subscribers.

You can see some of this in this video (forward to minute 2:12 if you just want to see the part about subscribing):


This is great Markus!

Now, are there any official plans to add this as native functionality to the reference agents?