Hermes FAQ
Straight answers to the questions people actually ask before trying Hermes: do I need to code, does it run on Windows, what does it cost, is my data private, and how do I start.
The honest answers to what people ask before they start. If your question is not here, the episode links point you to the full walkthrough.
Common questions
Do I need to know how to code?
No. You talk to Hermes in plain English, the same way you would chat with any AI assistant. The difference is that it can act on your computer, not just reply. If you can describe what you want, Hermes can usually do it. Coders get extra power from it, but it is not a requirement.
Does it work on Windows?
Yes, natively. There is a one-line PowerShell installer and no need for WSL or Linux. The installer even sets up the supporting tools for you. WSL2 is still fully supported if you prefer it.
Does it work on Mac and Linux?
Yes. On Mac and Linux you install it with a single curl command in your terminal, and it pulls in everything it needs in one shot.
Can I run it on my phone?
Yes, on Android through Termux, which gives you a Linux terminal on your phone. There is a tested install path for it. A few heavy extras are not available on Android, but the core agent runs.
How much does it cost?
Hermes itself is free and open-source under the MIT license. You only pay for the AI model usage from whichever provider you choose, and running a local model on your own machine is completely free. So your cost is whatever the model behind it costs, which can be zero.
Which provider should I start with, and do I need an API key?
The simplest start is Nous Portal: one sign-in gives you hundreds of models plus web search, images, voice, and a browser, with no separate keys to juggle. If you would rather bring your own provider like OpenRouter or OpenAI, you paste in an API key once and Hermes saves it. You can change providers any time.
Can I use it offline or with local models?
Yes. Point Hermes at a local server like Ollama, vLLM, or llama.cpp and it runs against a model on your own machine. That keeps everything private and free, at the cost of needing a capable computer.
Is my data private? Does it get sent anywhere?
Your messages go only to the AI provider you configure, and nowhere else. Hermes collects no telemetry, usage data, or analytics. Your conversations, memory, and skills are stored locally on your machine in the ~/.hermes folder. If you use a local model, nothing leaves your computer at all.
Will it change my files or run risky commands without asking?
No. Hermes snapshots your files before it edits them, so you can undo a change with /rollback. Potentially destructive commands are flagged and require your approval before they run. You stay in control of anything irreversible.
What is the difference between memory and skills?
Memory stores facts: things the agent knows about you, your projects, and your preferences. Skills store procedures: step-by-step instructions for how to do a task. Both stick around between sessions, but one is what it knows and the other is how it works.
Can I message it from my phone like a normal chat app?
Yes. The messaging gateway puts your agent on Telegram, Discord, Slack, WhatsApp, Signal, and more. You set it up once, run it, and then text it, send a voice memo, or share a file from anywhere.
Can more than one person use the same agent?
Yes. Through the messaging gateway, several people can talk to one Hermes instance. You control who gets in with an allowlist of approved users, or with DM pairing where the first person to message claims access.
How does it remember things between conversations?
Hermes keeps a small, curated set of notes in MEMORY.md and USER.md and loads them at the start of every session. It updates them on its own as it learns your setup, so you stop repeating yourself. The notes are kept deliberately compact so they stay useful.
Can it run tasks on a schedule while I am away?
Yes. Ask it in plain language to do something every morning or every hour, and it schedules the job and delivers the result to you. This is how people set up daily briefings and automated checks.
Can I use it inside my own Python project?
Yes. You can import the agent as a Python library and call it from your own code, or expose it as an OpenAI-compatible API server that other apps talk to. Both let you build Hermes into something larger.