Hermes Agent. Open source, MIT licensed, 140K GitHub stars and growing fast. Built by Nous Research.
This is the full breakdown. The mental model, the setup, the API key handling, the first cron worth running, the dashboard, the scaling rules, and every piece of practical advice I'd give you if we were sitting down to build one together.
## TL;DR
→ Hermes is an open source agent that grows with you. VPS, Mac Mini, laptop, Docker, even Android via Termux.
→ Five pillars: Memory, Skills, Soul, Crons, Self-Improving Loop.
→ Connects to Telegram, Discord, Slack, WhatsApp, iMessage.
→ I don't replace Claude Code with it. Hermes is the on-the-go, voice-first, scheduled-automation layer that lives in your pocket.
## What Hermes Actually Is
It's the agent in your pocket.
Claude Code is my daily driver for knowledge work and coding at my desk.
Hermes is what I talk to from Telegram while I'm walking the dog, on a flight, or away from the laptop.
Same brain, different interface.
Out of the box, 91 skills ship with it. The community hub has 520+ more. 16 of those are official Anthropic skills. I've never had to install Excalidraw or a transcription skill. Both were already there.
My main one runs:
→ A daily AI news briefing posted to my Skool community
→ YouTube comment monitoring with sarcastic-but-not-rude replies
→ Skool community engagement
→ Morning business summaries
→ Server health checks
→ Research reports
Generated by Thread Navigator
Press ⌘ + S to quick-export