Hermes is not difficult to install, but first distinguish the platform: it officially supports Linux, macOS, and WSL2, and Windows does not natively support it. If you are using it on Windows, install WSL2 first, and then execute the official one-click installation script; After installation, run hermes setup to select the model provider, and finally enter the hermes directly to start chatting.
The most hassle-free way to install
- Execute the official installation command in the terminal:
curl -fsSL https://raw.githubusercontent.com/NousResearch/hermes-agent/main/scripts/install.sh | bash - Complete the installation as prompted, and the script automatically handles uv, Python 3.11, repository clone, and global commands.
- Run the
hermes setupafter reloading the shell and configure Nous Portal, OpenRouter, or custom API endpoints.
If you want to go one step faster, you can also use hermes model to select the model directly; If the configuration is changed later, running hermes config check can quickly check for missing items. Many people get stuck, but most of them are because the shell is not refreshed or the provider is not fully filled.
How to get started
The most basic use is to enter hermes in the terminal, which runs as a fully interactive CLI that supports tools, memory, and skills. If you want it to connect to Telegram, Discord, Slack, WhatsApp, Signal, or Email, you can do hermes setup gateway again, turn on the gateway, and then configure the channel.
If you just want to "put it on and chat", it's enough to remember this order: WSL2 or Linux/macOS, install the script, hermes setup, then hermes. If you want to connect it to your workflow, open the gateway and configure the provider. At its core, Hermes is not just chatting, but putting memory, automation, and multi-platform agents into the same runtime.
