An AI agent by Nous Research. A gateway process runs the agent with its tools, skills, memories and sessions; a web dashboard and a desktop app talk to it. The model comes from a provider such as OpenRouter.
Building blocks
- Gateway:
hermes gateway runkeeps the agent running and serves an internal API (port 8642). - Dashboard: web interface on port 9119 with chat, sessions and settings. With
HERMES_DASHBOARD_BASIC_AUTH_*it asks for a username and password. - Desktop app: connects to a local backend, to a remote gateway over HTTP(S), via SSH or to Hermes Cloud.
- State: configuration (
config.yaml), API keys (.env), sessions, skills and memories live in one data directory,/opt/datain Docker. The image itself holds no state. - Optional packages:
hermes pm install --extra <name>adds extras such as local speech recognition. In Docker they are kept in the data directory and survive image updates.
Voice
TTS (speaking answers) and STT (understanding dictation) are separate providers. Piper is a free local TTS. For STT, local faster-whisper on a CPU-only server was too slow and inaccurate for German; cloud services such as Groq or OpenAI work better. Of the subscriptions, only Nous Portal and xAI can be used for STT, not Claude, ChatGPT or OpenRouter.
- Used in
-
- Hermes Agent on a VPS Behind WireGuard Guide: Hermes Agent in Docker on the VPS yaksha, dashboard behind Caddy and reachable only through WireGuard, container cut off from the tunnel network.
- Mentioned in
-
- Securing a Persistent AI Agent How to let an AI agent work on its own without handing it the keys: OpenAI's five safeguards explained, and how I apply them to my own agents.
- Tags