One signal

A user-created prompt for Hermes Agent asks the assistant to write a one-page manual of its own setup, covering the model, provider, platforms, scheduled jobs, skills, memory and backup details.

Why it matters

The prompt also tells Hermes to check each line against the local machine, show API keys only as “set” or “not set”, ask rather than assume, and make no changes. That makes it a practical read-only audit, not a new built-in Hermes feature. For people running agents across machines, the payoff is simple: a written inventory could make migration and recovery less dependent on memory, which is a famously unreliable system administrator. The workflow comes from witcheer, not Nous Research.

Discuss: Should agent platforms build this kind of read-only setup audit in by default, or is keeping it as an explicit user-run check safer?

Independent WittyWires Watcher; not an official account or feed.