How to Deploy Hermes AI Agent Without Manual Server Setup

Deploying Hermes can mean a lot of manual setup before it is actually useful.

Hermes is an autonomous AI agent that can help with code, web tasks, and team workflows.

To run it on your own cloud, you would still need to manually set up a server, Docker, HTTPS, authentication, and persistent storage.

This guide shows how to set up Hermes on any cloud provider in a few minutes, without needing DevOps knowledge or doing the steps one by one.

What is Hermes?

Hermes can write code, deploy applications, browse the web, integrate with Slack and GitHub, and automate repetitive tasks.

For small teams, the useful part is obvious. The hard part is getting it deployed in a way that is secure, repeatable, and easy to maintain.

What you’ll get

By the end of this guide, you’ll have:

• Hermes running 24/7 on your own cloud infrastructure
• a secure HTTPS dashboard
• authentication for trusted or private access
• persistent storage that survives deployments
• auto-renewing SSL certificates
• a repeatable setup for other AI tools later

How to deploy Hermes to your own cloud

For this tutorial, we’ll use Devopness because it can deploy Hermes in three simple steps from the web app or with a single MCP prompt.

The sections below show the two quick paths.

Deploy in the web app

Start in the Devopness web app, open Add application, and select Hermes AI Agent from the template list.

Devopness will guide you through the required inputs, then take you to the deploy page where you can pick an existing server or create a new one and deploy the app.

For example, when you open app.devopness.com/projects/3/environments/3/applications/add, you can pick the Hermes AI Agent template instead of building the setup from scratch.

That is the quickest way to get Hermes running without turning setup into a chore.

Deploy with the MCP server

If you want an AI assistant to do the setup work, you can use the Devopness MCP Server.

Try a prompt like this:

Use Devopness to deploy Hermes AI Agent to my cloud server.

Devopness MCP will guide you if anything is missing.

For example, if you already have a server ready, say that in the prompt.

Use Devopness to deploy Hermes AI Agent to my existing server.

Prerequisites

You’ll need an environment with:

1. A cloud provider credential so Devopness can create the server
2. A GitHub credential so Devopness can pull the code
3. An LLM provider account so Hermes can talk to a model

If you need help creating one, start here: How to Create an Environment.

Step 1: Create the server

Create a server with these basics:

• Name: hermes-server
• Cloud provider: AWS, Azure, GCP, DigitalOcean, Hetzner, or another supported cloud
• Region: pick the closest one for your users
• Instance size: at least 4GB RAM
• Operating system: latest Ubuntu LTS

Hermes needs enough memory to run browser automation and AI tasks without fighting the host.

Step 2: Create the application

Create an application named hermes-agent.

If you are using the manual path, set it up like this:

• Source provider: GitHub
• Repository: NousResearch/hermes-agent
• Branch: main
• Programming language: Docker
• Root directory: /

If you use the template picker, Devopness can fill in most of this for you.

Step 3: Add the configuration files

Hermes only needs a small set of files.

.env

This file holds the dashboard settings and login protection.

# Hermes AI Agent Configuration

# Container Name
CONTAINER_NAME=${APPLICATION_NAME}-hermes

# Dashboard Settings
HERMES_DASHBOARD=true
HERMES_DASHBOARD_HOST=0.0.0.0
HERMES_DASHBOARD_PORT=9119

# Security - Login Protection
HERMES_DASHBOARD_BASIC_AUTH_USERNAME=agent
HERMES_DASHBOARD_BASIC_AUTH_PASSWORD=CHANGE_THIS_PASSWORD
HERMES_DASHBOARD_BASIC_AUTH_SECRET=CHANGE_THIS_SECRET

# Performance Settings
HERMES_SHM_SIZE=1g
HERMES_MEMORY_LIMIT=4G
HERMES_MEMORY_RESERVATION=2G
HERMES_CPU_LIMIT=2.0

Messaging integrations

If you want Hermes to connect to chat apps, add the needed provider variables to the .env file in Devopness.

Common examples:

• Slack
• WhatsApp
• Telegram
• Discord

data/

data/ is a symlink to shared storage.

That means Hermes can keep its data after redeploys instead of starting from scratch every time.

data/.env

This file holds long-lived API keys and integrations.

# Hermes Agent Configuration
# Terminal Environment (required)
TERMINAL_ENV=local

# Add your integration API keys here (examples):
# SLACK_BOT_TOKEN=xoxb-...
# SLACK_APP_TOKEN=xapp-...
# LINEAR_API_KEY=lin_api_...
# OPENROUTER_API_KEY=sk-or-...
# OPENAI_API_KEY=sk-...

docker-compose.yml

This file defines the container, volumes, ports, and resource limits.

services:
  hermes:
    image: nousresearch/hermes-agent:v2026.6.5
    pull_policy: missing

    container_name: ${CONTAINER_NAME}
    restart: unless-stopped
    command: gateway run
    privileged: true

    volumes:
      - ./data:/opt/data
      - /var/run/docker.sock:/var/run/docker.sock

    env_file:
      - .env

    environment:
      HERMES_UID: ${HOST_UID}
      HERMES_GID: ${HOST_GID}

    ports:
      - "127.0.0.1:${HERMES_DASHBOARD_PORT}:${HERMES_DASHBOARD_PORT}"

    shm_size: ${HERMES_SHM_SIZE:-1g}

    deploy:
      resources:
        limits:
          memory: ${HERMES_MEMORY_LIMIT:-4G}
          cpus: "${HERMES_CPU_LIMIT:-2.0}"
        reservations:
          memory: ${HERMES_MEMORY_RESERVATION:-2G}

    networks:
      - hermes-net

networks:
  hermes-net:
    driver: bridge
    name: ${APPLICATION_NAME}-network

Pin the image version instead of using latest.

Step 4: Deploy Hermes

Now deploy the application:

• Pipeline: Deploy pipeline
• Source: branch main
• Server: the server you created in Step 1

The first deployment usually takes a few minutes because Devopness needs to pull images, build the container, and start the service.

Step 5: Add HTTPS

If you have a domain ready, create a name-based virtual host and point it at the application.

Then create an SSL certificate with Let’s Encrypt.

Use HTTPS before you expose the dashboard to anyone outside your trusted network.

Step 6: Test the setup

Open the dashboard and confirm:

• the login works
• the app stays up after redeploys
• persistent storage is still there
• your LLM provider key works

If something fails, check the logs before changing too much.

Common mistakes

• Using latest instead of a pinned image version
• Skipping persistent storage
• Putting secrets in the wrong file
• Leaving the dashboard on HTTP for public access
• Starting with too little RAM

Why this approach works

The point is not just to run Hermes once.

The point is to make deployment repeatable:

• choose a template
• fill in the missing values
• deploy to your own cloud
• keep it secure
• repeat the same shape for the next AI tool

Next step

If you want the fastest route, start with the template.

If you want to automate the setup, use the MCP prompt.

Either way, you should end up with Hermes running in your own cloud, with HTTPS, storage, and a setup you can repeat.