I ran our first Intermediate Hermes Agent workshop today with The AI Capitol and the community. Five hours and we covered more than just installing an agent. A few participants even joined us from overseas, converging in Singapore for the day.
We installed Hermes Agent inside a Docker sandbox rather than straight onto the machine, then connected it to Telegram so you can drive it from your phone. We set up a private VPN so the dashboard is accessible anywhere, while your data and your keys stay on your own box, not on the cloud.
We spent real time on safeguarding the agent. An agent is only as safe as the guardrails you give it, so we set up SOUL.md as the boundary file: what the agent can touch, what it must never do, and where it stops and asks. Permissions, approvals and memory limits are the difference between a useful agent and a liability. We also went in depth on local AI models.
Then we put it to work: set up workflows, orchestrated multiple Hermes agents, and supercharged the setup with reusable skills and tools so the same instructions run again and again.
Thank you to everyone who joined us today, and to The AI Capitol for making it happen.