Action Space: MCP Servers

My main agent, Tom, assists me to balance work and life responsibilities with the available actions defined in his "action space". This translates to building lite agents specialized at a task - much like daemon processors in Mac OS. I believe that the difference here is that each of these sub agents are from external sources so to enable them to communicate with one another is to attach their servers under shared network or subnets.

This is particularly important in not just securing personal data but also ensures that my code base is scalable - if my first Tom is not behaving accordingly, I can opt to build another Tom.

Not a huge fan of cybersecurity related modules but I have tinkered enough with dockers containers that allowed me to find ease in setting this up.

The end goal is to ensure the security of personal data and to persistTom's state in the same manner as https://learn.microsoft.com/en-us/power-bi/connect-data/service-gateway-onprem - one way of leveraging internet gatways to store data rather than actual databases. If this build is successful, he can never die - unless he chooses to.

Some bookmarked resources:

Last updated