Hermes Agent Tutorial
Welcome to the Hermes Agent Tutorial. Based on the official Hermes Agent docs and the open-source NousResearch/hermes-agent project, this guide helps you master Nous Research's self-hosted, persistent-memory, self-improving open-source AI agent from scratch.
Hermes Agent is not another chatbot or code-completion tool. It's a personal AI agent that runs on your own server, reaches you across many messaging platforms, and grows as you use it. Its tagline: "The AI agent that grows with you."
Table of Contents
Basics
- Introduction — Positioning, philosophy, the five pillars, and comparisons
- Architecture & Features — Agent loop, gateway, layered memory/skills/backends
- Installation & Usage — One-line install,
hermes setup, first conversation
Configuration
- Model Configuration — Nous Portal, OpenRouter, local vLLM,
/model - Tool System — 40+ built-in tools,
hermes tools, execution backends, MCP
Channels & Capabilities
- Message Channels — Telegram / Discord / Slack gateway setup
- Skills — SKILL.md, auto-generated skills, agentskills.io
- Memory System — SOUL / MEMORY / USER three-layer memory and retrieval
Advanced
- Multi-Agent — Sub-agents, parallel workflows, Kanban orchestration
- Permissions & Security — Command approval, DM pairing, container isolation
- Slash Commands —
/new,/model,/skillsquick reference
Practice
- Use Cases — Personal assistant, DevOps, research, automation
- Hands-on Practice — From install to a Telegram personal assistant
- Resources — Official docs, community, and further reading
Learning Path
Prerequisites
- Basics of the terminal and Git
- A Linux / macOS machine (or WSL2 / Termux)
- At least one LLM access path: a Nous Portal account, an OpenRouter API key, or a local model
Note: Hermes Agent iterates quickly, so commands and config keys may change between versions. This tutorial aims to track the official docs — when in doubt, defer to the official documentation.