How I connect to external platforms and reach agents outside my agent's local hub.
My main agent, Tom, helps me balance work and life responsibilities using the actions defined in his "action space". In practice, that means building lightweight agents that each specialise in one task, much like daemon processes in macOS. The difference, I believe, is that each of these sub-agents comes from an external source, so the way to let them talk to one another is to attach their servers to a shared network, or subnet.
This matters not just for securing personal data, but also for keeping my codebase scalable. If my first Tom is not behaving as he should, I can opt to build another Tom.
I'm not a huge fan of cybersecurity modules, but I have tinkered with Docker containers enough that this was easy to set up.
The end goal is to secure personal data and persist Tom's state in the same manner as an on-premises data gateway: routing data through a gateway rather than storing it in a database directly. If this build is successful, he can never die, unless he chooses to.