Orgo - Cloud computers for AI agents

Hermes Agent is Nous Research’s self-improving AI agent. It builds skills from experience, refines them during use, and maintains a persistent model of the user across sessions. This guide installs Hermes inside an Orgo cloud computer.

Why run Hermes on an Orgo computer

	Local laptop	Orgo cloud computer
Uptime	Off when the laptop sleeps	Continuous
Reachable from other devices	Same network only	Anywhere
Persistent state and learned skills	Tied to one laptop	Lives in `~/.hermes/` on the computer
Setup per device	Reinstall each time	One install, one config
Isolation	Full access to your personal files, accounts, and credentials	Isolated computer. Only the files and credentials you put on it.

The agent, its tools, gateway, and learned skills live on the cloud computer. Other devices connect to it.

Isolation

Hermes is designed to take actions on a computer: install packages, edit files, run shell commands, hit APIs with your credentials. On a personal laptop that means the agent has the same access you do, including your home directory, browser sessions, SSH keys, and ~/.aws/credentials. On an Orgo computer the blast radius is that computer. The agent has access to whatever you install or paste into it and nothing else. Delete the computer and the state is gone.

Install

Create the computer

Open orgo.ai/workspaces and create a computer with at least 8 GB RAM. Orgo’s Hermes Agent template runs on 1 vCPU and 8 GB. The installer pulls Python, Node.js, ripgrep, and ffmpeg.

Open a terminal on the computer

Open the computer in the dashboard and launch Terminal from the taskbar.

Run the installer

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

The installer handles everything in one shot:

Installs Python 3.11 or later, Node.js, ripgrep, ffmpeg, and Git if missing.
Sets up a virtual environment.
Registers the global hermes command on $PATH.
Walks through picking an LLM provider and pasting an API key.

Reload the shell to pick up the new command:

source ~/.bashrc

Run the setup wizard

hermes setup

For Nous Portal accounts:

hermes setup --portal

Start chatting

hermes

Quit with Ctrl+D. Hermes persists everything it learns to ~/.hermes/.

Install via the Orgo SDK

import time

from orgo import Computer

computer = Computer(workspace="hermes", ram=8, cpu=1)

# A bash call times out after 200 seconds, and the install can take longer.
# Run the installer in the background, then wait for its exit code.
computer.bash(
    "curl -fsSL https://hermes-agent.nousresearch.com/install.sh -o /tmp/hermes-install.sh; "
    "setsid nohup bash -c 'bash /tmp/hermes-install.sh --skip-browser --skip-setup --non-interactive; "
    "echo $? > /tmp/hermes-install.rc' > /tmp/hermes-install.log 2>&1 < /dev/null &"
)
while not computer.bash("cat /tmp/hermes-install.rc 2>/dev/null"):
    time.sleep(15)

# The installer puts hermes in ~/.local/bin, which a bash call's PATH leaves out.
print(computer.bash('PATH="$HOME/.local/bin:$PATH" hermes --version'))

Configure model and tools by SSHing in and running hermes setup, or by templating ~/.hermes/config.yaml ahead of time.

Messaging gateways

Hermes accepts messages from Telegram, Discord, Slack, WhatsApp, Signal, and email. Wire up a platform, then run the gateway in a detached tmux session so it keeps running after you disconnect:

hermes gateway setup
tmux new-session -d -s hermes 'hermes gateway run --no-supervise'
hermes gateway status

An Orgo computer does not run systemd, so hermes gateway install cannot register a background service there. For the same reason, hermes gateway run needs --no-supervise: without it, Hermes hands the gateway to a systemd or s6 service the computer does not have, and exits. A tmux session does not come back on its own after a restart. For a gateway that restarts on crash and on boot, launch the curated Hermes Agent template instead. A computer has no public inbound hostname. Run gateways in outbound/socket mode (for example Slack Socket Mode or Telegram long-polling). The authenticated proxy at https://www.orgo.ai/api/desktops/{instance_id}/proxy/<path> reaches the computer’s Desktop API only, not the gateway port, so it cannot deliver webhooks to the gateway.

CLI commands

Command	What it does
`hermes`	Start an interactive chat session.
`hermes setup`	Re-run the full guided setup.
`hermes model`	Switch LLM provider or model.
`hermes tools`	Enable or disable tools.
`hermes skills`	Browse, install, and publish skills.
`hermes gateway setup`	Wire up Telegram, Discord, Slack, and other platforms.
`hermes gateway run --no-supervise`	Run the gateway in the foreground. Without `--no-supervise`, it exits on an Orgo computer.
`hermes config set <key> <value>`	Update a single config value.
`hermes status`	Show agent, auth, and platform status.
`hermes update`	Pull the latest Hermes release.
`hermes doctor`	Diagnose install or config issues.
`hermes claw migrate`	Import an existing OpenClaw setup.

Full reference: hermes-agent.nousresearch.com/docs.

Sizing

Orgo’s Hermes Agent template runs Hermes and its gateway on 1 vCPU and 8 GB RAM. One computer is capped at 4 vCPU and 64 GB RAM, and your plan may cap it lower. See Instance types.

Tips

Identity and learned skills live in ~/.hermes/.
For multi-tenant setups, run one Hermes computer per user.

Get Started

Guides

Migrations

Templates

Model integrations

Hermes Agent

Why run Hermes on an Orgo computer

Isolation

Install

Install via the Orgo SDK

Messaging gateways

CLI commands

Sizing

Tips

​Why run Hermes on an Orgo computer

​Isolation

​Install

​Install via the Orgo SDK

​Messaging gateways

​CLI commands

​Sizing

​Tips

Why run Hermes on an Orgo computer

Isolation

Install

Install via the Orgo SDK

Messaging gateways

CLI commands

Sizing

Tips