How to Install Hermes Agent: Windows, macOS, Linux, and 24/7 Cloud Setup

AI Takeaway

• What is the fastest way to install Hermes Agent? Use the desktop installer on macOS or Windows, or run the official one-line command on Linux, macOS, WSL2, or Android Termux.
• What comes after installation? Reload your terminal, connect an AI model with hermes setup --portal or hermes model, run hermes doctor, and send one test prompt.
• Does Hermes Agent run natively on Windows? Yes. Windows 10 and 11 can use the native PowerShell installer. WSL2 is optional.
• Will it keep working when your laptop is closed? Not on a local installation. Scheduled jobs and messaging channels need an always-on computer, VPS, Docker host, or managed cloud instance.
• Which setup should you choose? Start locally if you want to explore. Use an always-on environment when Hermes needs to monitor, schedule, remember, and respond around the clock.

Choose How You Want to Run Hermes Agent

Hermes can run on your computer, a spare machine, a VPS, or a managed cloud instance. Choose based on what it must do after setup.

Method	Best for	What you manage
Desktop app	Trying Hermes on macOS or Windows	Your model, tools, and computer uptime
Local CLI	Development and local-file work	Runtime, updates, credentials, and uptime
Docker or VPS	A self-hosted agent that stays online	Server security, storage, backups, and recovery
Managed cloud	24/7 use without server maintenance	Models, channels, skills, and tasks

If you are still deciding whether you need a persistent general-purpose agent or a coding-focused tool, the Hermes Agent vs Claude Code comparison clarifies the difference.

Install Hermes Agent on Your Device

Use the current official installer. Hermes changes quickly, and older guides may still say native Windows is unsupported. On macOS or Windows, the desktop installer can set up the app and CLI together.

Install on macOS or Linux

Open Terminal and run:

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

The installer prepares the Hermes runtime and its main dependencies, including Python, Node.js, ripgrep, and ffmpeg. When it finishes, open a new terminal window. You can also reload the current shell:

source ~/.zshrc   # macOS and zsh
source ~/.bashrc  # most Linux shells

Confirm that the command is available:

hermes --version

On a headless server that will not use browser automation, you can skip the browser component:

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

Use this option only when you know the agent will not need browser tools.

Install on Windows 10 or 11

Open PowerShell or Windows Terminal and run:

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

Hermes now runs natively on Windows, so WSL2 is not required. The PowerShell installer uses a per-user location and adds the hermes command to your PATH. Close and reopen the terminal if PowerShell cannot find it immediately. If you prefer a graphical app, the desktop installer is the simpler option and includes the CLI runtime.

WSL2 remains useful when your workflow depends on Linux shell tools or you want the same environment as an Ubuntu server. Inside WSL2, follow the Linux installation command rather than the native PowerShell command.

Install on Android With Termux

Hermes can run inside Termux using the same one-line shell installer. Treat Android as an experimental, best-effort setup rather than a full server replacement. The CLI, cron support, Telegram gateway, memory, and core tools can work, but voice, browser automation, WhatsApp tooling, Docker isolation, and background persistence may be limited.

Android is a useful portable option, but less reliable through battery management, reboots, and network changes.

Skip the Setup, Run Best-in-Class AI Agents Now

Launch a managed OpenClaw or Hermes Agent workspace in minutes, with always-on hosting, updates, and support handled by MyClaw.

Get Started →

Finish the Setup and Confirm It Works

An installed command is only the first checkpoint. Hermes still needs a model provider before it can respond.

Connect an AI Model

For the shortest guided setup, run:

hermes setup --portal

To choose a different cloud service, compatible endpoint, or local model, use:

hermes model

Hermes itself is open source, but a cloud model may charge for API usage. Model quality, tool calling, speed, and price can differ substantially. The best AI models for Hermes Agent guide can help you choose without turning the installation into a long provider experiment.

Run a Health Check and Test Prompt

Check the environment before connecting extra tools:

hermes doctor
hermes chat -q "Reply with OK if Hermes is working."

You have a working base installation when the CLI starts, the model replies, and hermes doctor shows no critical issue. Add one layer at a time after that: tools first, then the Dashboard, messaging channels, skills, and scheduled jobs. If you want reusable workflows rather than repeated prompts, the Hermes Agent skills guide explains how skills capture a process and improve it over time.

Fix Common Installation Problems

Most setup failures become easier once you identify which layer is broken.

Problem	Likely cause	What to do
hermes: command not found	The terminal has not loaded the new PATH	Open a new terminal or reload the shell profile
Provider or API-key error	Model setup was skipped or credentials are misplaced	Run hermes model again and confirm the selected provider
Browser tool fails	Browser dependencies were skipped or are unavailable	Reinstall the browser component and rerun hermes doctor
CLI works but Telegram or Discord does not	The messaging gateway is a separate process	Configure the gateway and test a real incoming message
Memory or skills disappear after a Docker restart	Hermes data was stored inside the temporary container	Mount the Hermes data directory on persistent storage

Debug in this order: installation → model → tools → terminal backend → gateway. Do not change several layers at once. A successful CLI response proves that the runtime and model are healthy before channel configuration enters the picture.

Keep Hermes Running After Installation

A local agent stops when its computer sleeps. For daily briefs, monitoring, Telegram replies, or scheduled work, move Hermes to a machine that stays online.

Preserve Memory, Skills, and Settings

The ~/.hermes directory contains the state that makes the agent yours: configuration, profiles, sessions, memory, and skills. Native Windows stores its corresponding data under %LOCALAPPDATA%\hermes.

Mount this state on persistent storage when using Docker. On a VPS, back it up before upgrades and test that the backup can actually be restored. A backup that has never been restored is only a hopeful copy.

Secure the Dashboard and Gateway

Do not expose an unauthenticated Dashboard directly to the public internet. Use authentication, HTTPS or private networking, limited credentials, and only the ports you need. Add process supervision or a restart policy so the gateway returns after a crash or reboot.

If you are comparing runtimes before committing to long-term hosting, OpenClaw vs Hermes Agent covers the practical differences in control, memory, and workflow design.

Run Hermes Agent 24/7 With MyClaw

Running a VPS is reasonable when you enjoy managing the infrastructure. If you want Hermes to stay online without maintaining Linux, Docker volumes, restart services, updates, and backup jobs, MyClaw provides a hosted Hermes workspace with those operational pieces handled for you.

You still decide what the agent does. MyClaw keeps the runtime online and brings its WebUI, channels, skills, cron jobs, updates, backups, and recovery controls into one place.

Step 1: Choose a Workspace

Start with the workload, not the largest server. A personal assistant that sends one daily briefing needs less capacity than several profiles handling research, code, and monitoring at the same time.

Get Started

Step 2: Connect the Model and Channels

Launch the Hermes instance, connect your preferred model provider or add your own API key, and enable only the channels needed for the first workflow. Beginning with one model and one channel keeps testing simple.

Step 3: Give Hermes a Job Worth Keeping Online

Pick a task that benefits from continuity: deliver a morning market brief, watch a product page for changes, summarize incoming messages, or send a weekly report to Telegram or Slack. Once that works reliably, add the next workflow instead of configuring everything on day one.

Install Hermes Agent and Put It to Work

A local Hermes Agent install is the quickest way to understand how the system feels. Complete the model setup, run the health check, and give it one useful task. The full Hermes Agent review offers a closer look at memory, skills, and automation before you expand the setup. When your first task needs remote channels or a schedule that keeps running after your laptop closes, move the same workflow to an always-on environment.