R Rumus Cloud
Course · 2 days

Hermes and Claude on your own server

Run an agent that works from your own infrastructure, with Claude as its model and your data staying where it is. Two sessions, ending with a working deployment.

What you leave with

  • Hermes running on your own server, with Claude as the configured model
  • MCP servers connected for the tools you actually use
  • Scheduled jobs that report somewhere your team will read
  • Keys kept out of shell history and out of the repository
  • A way to let colleagues use it without giving them the server

Day one

BlockContent
09.00–10.30Installation, provider configuration, first conversation
10.45–12.00Model selection and cost behaviour: what a job actually costs
13.00–14.30MCP: connecting tools, reading what the agent can reach
14.45–16.00Practice: a job that checks something real and reports it

Day two

BlockContent
09.00–10.30Schedules, webhooks and delivery to chat
10.45–12.00Giving the team access without handing over root
13.00–14.30Guardrails: destructive actions, approvals, audit trail
14.45–16.00Bring your own use case, built with assistance
Bring a real problem. The second afternoon works far better when you bring a task your team repeats weekly. People who arrive with one leave with something in use.