- Any Linux, macOS, or Windows machine (yes, native Windows is fully supported) - A $5 VPS works fine. A GPU cluster also works. Hermes scales to what you have. - Storage: ~2GB for the install (agent code + dependencies + model cache) - RAM: 2GB minimum for basic use; 8GB+ recommended for running local models The installer handles almost everything automatically: | Requirement | Notes | |-------------|-------| | Git | Must be installed (git --version to check) | | curl + xz-utils | (Linux only)

Hermes Agent Beginner's Guide: Install and Run Your First Self-Improving AI Agent

Q: What Is Hermes Agent?

Hermes Agent is an open-source, self-improving AI agent built by Nous Research] — the team behind some of the most influential open-weight models in AI. Released in February 2026 under the MIT license, Hermes has quickly become one of the most talked-about open-source AI projects of the year. What makes it different from everything else: | Feature | What it means | |---------|---------------| | Persistent memory | Remembers your preferences, projects, and context across sessions. No more re-expl

Q: Who Is This Guide For?

- Developers who want a personal AI agent on their own infrastructure - AI enthusiasts curious about open-source agent frameworks - MLOps engineers looking for a production-grade agent harness - Anyone who's tired of chatbots that forget everything between conversations You don't need deep AI knowledge. You do need basic terminal familiarity. ---

Q: Step 1: Install Hermes Agent

curl -fsSL https://hermes-agent.nousresearch.com/install.sh | bash That's it. The installer: 1. Downloads and installs uv (a fast Python package manager) 2. Installs Python 3.11 (via uv, no system Python change) 3. Clones the Hermes Agent repository 4. Sets up a virtual environment with all dependencies 5. Creates the hermes command (symlinked to ~/.local/bin/hermes) 6. Installs Node.js v22, ripgrep, and ffmpeg if missing After installation, reload your shell: source ~/.bashrc or source ~/.zshrc

Q: Step 2: Choose Your AI Provider

The most important decision. Run: hermes model This launches an interactive setup wizard. Here are your best options: hermes setup --portal This single command logs you in via OAuth, sets Nous Portal as your provider, and enables the Tool Gateway (web search, image generation, TTS, cloud browser) — all with zero API key management. Nous Portal gives you access to 300+ models under one subscription. No juggling API keys across providers. | Provider | How to Set Up | Best For | |----------|-------

"Not a chatbot. Not a code completion tool. An AI agent that lives on your server and gets smarter the longer you run it."

What Is Hermes Agent?

Hermes Agent is an open-source, self-improving AI agent built by Nous Research — the team behind some of the most influential open-weight models in AI. Released in February 2026 under the MIT license, Hermes has quickly become one of the most talked-about open-source AI projects of the year.

What makes it different from everything else:

Feature	What it means
Persistent memory	Remembers your preferences, projects, and context across sessions. No more re-explaining everything each time.
Automatic skill creation	When Hermes solves a hard problem, it writes a reusable skill document so it never forgets the solution.
Multi-platform	One agent, accessible from Telegram, Discord, Slack, WhatsApp, Signal, and CLI — all through a single gateway.
Self-hosted	Runs on your own server. All data stays in `~/.hermes/`. No telemetry, no tracking, no cloud lock-in.
Any model	Swap between 300+ models from Nous Portal, OpenAI, Anthropic, Google, xAI, OpenRouter, local vLLM, and more — with a single command.
40+ built-in skills	Code execution, web search, browser automation, image generation, GitHub, file operations — everything works out of the box.

As of July 2026, Hermes Agent has over 17,000 GitHub stars and 207+ contributors.

Who Is This Guide For?

Developers who want a personal AI agent on their own infrastructure
AI enthusiasts curious about open-source agent frameworks
MLOps engineers looking for a production-grade agent harness
Anyone who's tired of chatbots that forget everything between conversations

You don't need deep AI knowledge. You do need basic terminal familiarity.

Prerequisites

Hardware

Any Linux, macOS, or Windows machine (yes, native Windows is fully supported)
A $5 VPS works fine. A GPU cluster also works. Hermes scales to what you have.
Storage: ~2GB for the install (agent code + dependencies + model cache)
RAM: 2GB minimum for basic use; 8GB+ recommended for running local models

Software Prerequisites

The installer handles almost everything automatically:

Requirement	Notes
Git	Must be installed (`git --version` to check)
curl + xz-utils	(Linux only) Needed for Node.js download
build-essential	(Linux, for desktop app) `sudo apt install build-essential`
Everything else	Python 3.11, Node.js v22, ripgrep, ffmpeg — all auto-installed

No sudo required for the standard install. The installer detects what's missing and handles it.

Step 1: Install Hermes Agent

The One-Liner (Linux / macOS / WSL2)

curl -fsSL https://hermes-agent.nousresearch.com/install.sh | bash

That's it. The installer:

Downloads and installs uv (a fast Python package manager)
Installs Python 3.11 (via uv, no system Python change)
Clones the Hermes Agent repository
Sets up a virtual environment with all dependencies
Creates the hermes command (symlinked to ~/.local/bin/hermes)
Installs Node.js v22, ripgrep, and ffmpeg if missing

After installation, reload your shell:

source ~/.bashrc   # or source ~/.zshrc

Windows (Native PowerShell)

iex (irm https://hermes-agent.nousresearch.com/install.ps1)

Native Windows is fully supported — CLI, gateway, TUI, and tools all work natively without WSL. The installer bundles a portable MinGit (~45MB) that lives entirely in %LOCALAPPDATA%\hermes\git.

Android (Termux)

If you're installing on a phone, see the dedicated Termux guide.

Desktop Installer (macOS / Windows)

For a graphical install including both CLI and desktop app, download from hermes-agent.nousresearch.com.

Step 2: Choose Your AI Provider

The most important decision. Run:

hermes model

This launches an interactive setup wizard. Here are your best options:

🚀 Fastest Path: Nous Portal (Recommended for Beginners)

hermes setup --portal

This single command logs you in via OAuth, sets Nous Portal as your provider, and enables the Tool Gateway (web search, image generation, TTS, cloud browser) — all with zero API key management.

Nous Portal gives you access to 300+ models under one subscription. No juggling API keys across providers.

Other Popular Options

Provider	How to Set Up	Best For
OpenAI Codex	`hermes model` → OAuth login	ChatGPT subscribers
Anthropic	`hermes model` → OAuth (Max plan)	Claude fans
OpenRouter	Enter API key	Multi-provider flexibility
Local vLLM	Custom endpoint	Privacy/cost control
GitHub Copilot	OAuth login	Copilot subscribers
xAI Grok	OAuth (SuperGrok)	Grok users
DeepSeek	Set DEEPSEEK_API_KEY	Cost-effective

Pro tip: Start with Nous Portal for the smoothest experience. You can switch providers anytime with hermes model.

Step 3: Verify It Works

After choosing your provider, start a conversation:

hermes

This launches the Hermes TUI (Terminal User Interface) — a full interactive environment with:

Multiline editing (write complex prompts across multiple lines)
Slash-command autocomplete (type / to see available commands)
Conversation history (scroll through past exchanges)
Streaming tool output (watch the agent think in real time)

Try these prompts:

What skills do you have available?

Can you search the web for the latest AI news?

Write a Python script that fetches today's top HN stories.

If Hermes responds, you're up and running. If not, see the troubleshooting section below.

Exiting the TUI

Press Ctrl+C or Ctrl+D to exit the conversation.

Step 4: Essential Commands

Once you have a working agent, these commands will be your daily toolkit:

Command	What it does
`hermes`	Start a TUI conversation
`hermes model`	Change your AI provider/model
`hermes tools`	Enable/disable tools (web, browser, code, etc.)
`hermes setup`	Run the full setup wizard
`hermes update`	Pull latest code and reinstall dependencies
`hermes config set <key> <value>`	Set individual config values
`hermes gateway setup`	Configure messaging platforms (Telegram, Discord, etc.)
`hermes skills`	List installed skills
`hermes --help`	Show all available commands

Understanding Your Agent's Home

Everything Hermes knows lives in ~/.hermes/:

~/.hermes/
├── hermes-agent/     # The cloned repository
├── config.yaml       # Your settings (provider, tools, etc.)
├── sessions/         # Conversation history
├── memory/           # Long-term memory store
├── skills/           # Auto-created and installed skills
└── data/             # Agent-local data files

Privacy note: All data stays in this directory on your machine. No telemetry, no phoning home, no cloud dependency. You own everything.

What Can You Do Now?

With a working Hermes Agent, you can immediately:

✅ Have persistent conversations that remember context across sessions ✅ Search the web and browse pages ✅ Execute Python and shell commands (with safety sandboxing) ✅ Run code analysis and debugging ✅ Access your file system (read/write with permission) ✅ Generate images (if your provider supports it)

Troubleshooting

"hermes: command not found" after install

source ~/.bashrc   # or source ~/.zshrc

Installation fails on Linux

Ensure prerequisites are installed:

sudo apt install curl xz-utils git

Model doesn't respond

Check your provider config: hermes model
Make sure you have sufficient credits/quota
Try Nous Portal for a zero-config setup: hermes setup --portal

TUI seems stuck

Wait for the agent to finish thinking (it streams output)
Press Ctrl+C to interrupt and get back to the prompt
Ensure your terminal supports UTF-8 and true color

What's Next?

This is Phase 1 of our Hermes Agent series. You've installed your agent, run your first conversation, and learned the CLI basics. Here's what's coming:

Phase	Title	Covers
🟢 Phase 1	Beginner's Guide ← You're here	Install, configure, first conversation, CLI basics
🟡 Phase 2	Making Hermes Remember	Memory system, skill creation, multi-platform gateway (Telegram/Discord/Slack)
🟠 Phase 3	Advanced Operations	Custom models, automation (cron jobs), parallel sub-agents, custom tools
🔴 Phase 4	The Master Level	RL training, trajectory generation, fine-tuning your own agent models

Your mission for Phase 1: Get Hermes running, have at least three conversations with it, and try asking it to create a skill. See what happens when an agent that learns meets your real workflow.

Resources

Official site: hermes-agent.nousresearch.com
GitHub: github.com/NousResearch/hermes-agent
Documentation: hermes-agent.nousresearch.com/docs
Discord: Nous Research Discord
Video tutorials: Hermes Agent Tutorials & Use Cases (YouTube)

Find this guide helpful? Share it with someone who's curious about running their own AI agent.

Update frequency: This guide is accurate as of July 9, 2026 (Hermes Agent latest stable release).