Deploy Hermes Agent

Hermes Agent is a self-hosted AI agent from Nous Research. It runs on your own server, works with files and tools, executes terminal commands, and keeps a history of your sessions. You interact with the agent through a text-based interface in the terminal.

Hostman provides a ready-to-use image with Hermes pre-installed. Once you've created a server, all that's left is to connect it to an AI agent or to AI Gateway.

Creating a Server

Copy link

1. In the control panel, go to the Cloud Servers section.
2. Click Create Server.
3. Choose a region for your server.
4. On the Image step, open the Marketplace tab and select the Hermes image.

5. Choose a server configuration and adjust any other settings as needed (you can leave the defaults).
6. Click Create.

Hermes and all the components it needs are installed automatically when the server is created.

Choosing a Connection Method

Copy link

You can connect Hermes to either an AI agent or to AI Gateway.

If you connect to an AI agent, you'll be able to use a knowledge base. In this case, Hermes works with a single model—whichever one is selected in the agent's settings. To switch models, change the model in the Hostman dashboard.

If you connect to the AI Gateway, you won't be able to use AI agent knowledge bases. However, you'll be able to choose any model available in AI Gateway and switch between models on the fly.

Before you start, gather the connection details you'll need:

For an AI agent

• Base URL of the agent: shown in the agent’s Dashboard
• Access key

If you haven't created an AI agent yet, follow this guide.

Agent's Base URL in the Dashboard

For AI Gateway

• Base URL: https://ai-api.hostman.com/v1
• AI Gateway API key

Note that AI Gateway uses its own separate keys, unrelated to AI agent keys.

Setting Up the Connection

Copy link

Step 1. Connect to your server over SSH

1. Open a terminal: Terminal on macOS/Linux, PowerShell on Windows.

2. Open the server's Dashboard tab and copy the command from SSH connection.

SSH connection command in the Dashboard

3. Paste the command into your terminal and press Enter.

Step 2. Confirm the connection (first time only)

On your first connection, you'll see a prompt like this:

Are you sure you want to continue connecting (yes/no/[fingerprint])?

Type yes and press Enter.

Enter the root password from the Dashboard when prompted. The password won't be displayed as you type—this is normal terminal behavior.

Server root password in the Dashboard

Step 3. Exit Hermes to access the server console

The Hermes text interface opens automatically once you connect. 

To run the setup script in the next step, you need to return to the server console first.

To do this, type /exit and press Enter.

Step 4. Run the setup script

Run this command in the terminal:

curl -fsSL https://st.hostman.com/cloud-static/hostman-hermes-autoinstall.sh -o /tmp/hostman-hermes-autoinstall.sh && bash /tmp/hostman-hermes-autoinstall.sh

Step 5. Enter your configuration details

The script will prompt you for:

• Base URL: The base URL of your AI agent or AI Gateway
• API Key: The access key

Step 6. Review and confirm

After you enter these values, the script checks the connection and displays the resulting configuration.

Review the values. If everything looks correct, type y and press Enter.

The script will add the ai-agent provider and open the chat interface.

Hermes Agent interface in Windows PowerShell

Selecting a Model in AI Gateway

Copy link

This step is only needed if you're connecting through AI Gateway. When using an AI agent, Hermes always talks to the model selected in the agent's settings.

In the Hermes interface, run:

/model

This opens the model selection window. Choose the ai-agent provider, then select the model you want to use.

You can change the model at any time during your session using the same /model command.

Working with Hermes

Copy link

To manually launch the Hermes text interface, run:

hermes --tui

From this interface, you can send requests to the agent, view its responses, and track tool executions.

Useful commands:

To resume your last session after exiting, run:

hermes --tui --continue

To open a specific saved session by its ID or name, use:

hermes --tui --resume "<id_or_name>"

Key keyboard shortcuts:

Once everything is set up, send the agent a test request to confirm it responds correctly. If you're using AI Gateway, select a model with /model first.

For other commands and interface features, see the official Hermes documentation.