Hermes Agent OS Support: macOS, Linux, Windows, Android

Supported Operating Systems

Per the upstream Hermes Agent installation docs, these operating systems are first-class targets:

System Requirements

• Disk: ~2 GB for Hermes + dependencies; allow more if you cache local models
• RAM: 4 GB minimum; 8 GB+ recommended for smoother agent runs
• Network: Outbound HTTPS to your model provider (or a local model server if you go fully offline)
• Model context: A model with at least 64,000 tokens of context. Smaller windows can't hold the agent's working memory for multi-step tool-calling workflows.

The One-Liner Installer

On macOS, Linux, WSL2, and Termux you can install Hermes Agent in one command:

curl -fsSL https://raw.githubusercontent.com/NousResearch/hermes-agent/main/scripts/install.sh | bash

The installer handles everything automatically: Python, Node.js, ripgrep, ffmpeg, the repo clone, a virtual environment, and the global hermes command setup. After install finishes, run hermes with no arguments to enter interactive provider setup (Nous Portal, OpenRouter, Anthropic, OpenAI, Ollama, and others).

macOS Install

One-liner (recommended)

curl -fsSL https://raw.githubusercontent.com/NousResearch/hermes-agent/main/scripts/install.sh | bash

Works on Apple Silicon (M1/M2/M3/M4) and Intel Macs. The script installs into ~/.hermes-agent and adds hermes to your PATH via your shell rc file (~/.zshrc on modern macOS).

Verify

hermes --version
hermes # interactive setup

If you hit a permissions issue with Homebrew installing dependencies, the script will print which packages it needs — install them manually with brew install [email protected] node ripgrep ffmpeg and re-run.

Linux Install

One-liner

curl -fsSL https://raw.githubusercontent.com/NousResearch/hermes-agent/main/scripts/install.sh | bash

Tested on Ubuntu 22.04 / 24.04, Debian 12, Fedora 40+, and Arch. The installer detects your package manager (apt, dnf, pacman) and pulls dependencies automatically.

Docker alternative

If you'd rather not install Python and friends on the host, use the official Docker image:

docker run -d \
  --name hermes \
  -v ~/.hermes:/root/.hermes \
  -p 8642:8642 \
  ghcr.io/nousresearch/hermes-agent:v0.14.0  # pin to a tested release

See the Hermes Agent Docker guide for the production-grade Docker Compose setup with persistent volumes and reverse proxy.

Windows Install

WSL2 (recommended)

On Windows 10 or 11, install WSL2 with an Ubuntu distribution and run the standard one-liner inside it:

# In PowerShell (as Administrator):
wsl --install -d Ubuntu

# Then inside the Ubuntu WSL shell:
curl -fsSL https://raw.githubusercontent.com/NousResearch/hermes-agent/main/scripts/install.sh | bash

This is the most stable Windows path because Hermes was developed against Linux/macOS and the WSL2 environment matches that closely. See the Hermes on Windows guide for the full WSL2 setup walkthrough.

Native Windows (beta)

Native Windows support exists but is in early beta. Some skills that depend on POSIX shell behavior may not work. Install via PowerShell:

iwr -useb https://raw.githubusercontent.com/NousResearch/hermes-agent/main/scripts/install.ps1 | iex

If you hit unexpected behavior on native Windows, switch to WSL2 — it's the better-tested path.

Android Install (Termux)

Hermes Agent can run on Android phones and tablets via Termux, the terminal emulator that ships a real Linux environment. Useful for personal assistants you want to run on a phone overnight or on a Raspberry Pi-style portable setup.

1. Install Termux from F-Droid (the Play Store version is outdated and broken).
2. In Termux: pkg update && pkg upgrade
3. Install prerequisites: pkg install curl git python nodejs-lts ffmpeg ripgrep
4. Run the standard one-liner installer.

Storage and CPU on Android phones are tight; pick a small-context model and avoid heavy memory plugins (memory search embeddings can pin a phone's RAM).

After Install: Pick a Model Provider

When you run hermes for the first time it walks you through provider selection. Common picks:

• Nous Portal — one subscription, 300+ models, bundled tools (web search, image gen, TTS, browser)
• OpenRouter — one API key for 300+ models, granular spend control
• Anthropic / OpenAI / Google direct — direct provider keys, no gateway markup
• Ollama (local) — run a local model entirely offline

Whichever you pick, Hermes stores the auth in ~/.hermes/auth.json (or the platform equivalent). You can switch providers later with hermes setup.

Common Issues by OS

macOS: command not found: hermes after install

The installer adds Hermes to your PATH in ~/.zshrc. If you're running a non-zsh shell, source the rc file manually or run source ~/.zshrc in your current shell.

Linux: ffmpeg or ripgrep missing

Older distros sometimes don't have current versions in their default repos. Install via your package manager: sudo apt install ffmpeg ripgrep (Debian/Ubuntu), sudo dnf install ffmpeg ripgrep (Fedora).

Windows native: skills fail silently

Some skills rely on POSIX shell commands and don't work on native Windows yet. Switch to WSL2 if you hit this.

Termux: pkg install python takes forever

Termux repos can be slow from some regions. Try a different repo mirror (termux-change-repo) and re-run.

Don't Want to Run a Server?

If installing on your own machine isn't the right fit — you don't want to manage Python and Docker on every device — OpenClaw Launch's Hermes hosting runs Nous Hermes models on managed OpenClaw infrastructure today via OpenRouter, no OS install required. Native Hermes framework hosting (full Hermes Agent runtime with its memory and skill ecosystem) is in private beta — waitlist on the same page.

Related Guides

• Install Hermes Agent (full walkthrough)
• Hermes Agent on Windows (WSL2 setup)
• Hermes Agent with Docker
• What is Hermes Agent?
• Nous Portal subscription guide