Before Part 1: How to Install Hermes in 10 Minutes (Pre-Read)
This is the pre-read for the Agent Fleet Series. Part 3 onwards assumes you have Hermes running. Here is the shortest honest path to get there: install, pick a model, send one message, check it's healthy. About ten minutes.
What you need first
- A machine: Linux, macOS (Apple Silicon only) or Windows. WSL2 works too.
- Git, curl and tar on Linux, macOS or WSL2. On native Windows the installer can fetch Git for you.
- A model to talk to: an API key from a provider, or a Nous Portal subscription. Hermes works with most providers, so you are not locked in.
You do not need to install Python or Node yourself. The installer brings its own pinned versions.
Step 1: Install
On Linux, macOS or WSL2:
curl -fsSL https://hermes-agent.nousresearch.com/install.sh | bash
On native Windows, in PowerShell:
iex (irm https://hermes-agent.nousresearch.com/install.ps1)
If you would rather have an app than a terminal, macOS and Windows also have a desktop package on the Hermes website. If you install the command line first, you can open the desktop app later with hermes desktop.
The installer puts the code in ~/.hermes/hermes-agent/, a hermes launcher in ~/.local/bin/, and your data in ~/.hermes/. It also installs browser tools by default. Add --skip-browser to the install command if you don't want them.
Step 2: Reload your shell
source ~/.bashrc # or: source ~/.zshrc
Then confirm the launcher is found:
hermes --version
If it says command not found, add ~/.local/bin to your PATH.
Step 3: Pick a model
Run the setup wizard:
hermes setup
It walks you through provider and model. To change just the model later:
hermes model
The fastest route is one command that logs you in to Nous Portal and sets it as your provider:
hermes setup --portal
Where do the secrets go? API keys live in ~/.hermes/.env. Settings live in ~/.hermes/config.yaml. Keep that split: keys in .env, everything else in config.
Step 4: Say hello
hermes
That opens an interactive chat. For a single question without opening a session:
hermes chat -q "What can you help me with?"
If you get an answer, the install works.
Step 5: Run the health check
hermes doctor
It reports what's missing and how to fix it. Add --fix to let it repair what it can.
Settings you'll use from here
hermes config set <key> <value> # change one setting
hermes config get <key> # read one setting
hermes tools # choose which tools are enabled
hermes gateway setup # connect Telegram, Discord, Slack, WhatsApp...
hermes update # update Hermes itself
Use hermes config set rather than editing config.yaml by hand. One stray indent in a YAML file can break a running setup.
The prompt-based path
Once Hermes is running, you can ask it to check itself:
Check that my Hermes install is healthy. Run the built-in health check, tell
me anything that is missing or misconfigured, and tell me which model and
provider I am currently using. Don't change any settings without asking me.
Why this prompt works: it points at the built-in health check instead of letting the assistant improvise its own, it asks for the current model and provider so you learn where your setup stands, and "don't change anything without asking" keeps a diagnostic from turning into a rewrite.
Where this leads
Everything in the series builds on this one install. Part 3 turns it into an orchestrator profile with its own model and rules. Part 4 clones that profile into four specialists. If something goes wrong, hermes doctor is the first thing to run.
The official docs are at hermes-agent.nousresearch.com/docs if you hit a case this page doesn't cover.