Hermes Agent: what it is, how to use it and what it can do

What is Hermes Agent?

Hermes Agent is an open-source, MIT-licensed agent built by Nous Research and released in February 2026. It isn't a coding assistant tied to an editor. You run it on a laptop, a small server or serverless hosting, and talk to it in a terminal or through a messaging app.

Its main idea is that it gets more useful the longer it runs. It keeps memory between sessions, writes reusable skills from tasks it has finished, and can search its own past conversations.

What can Hermes Agent do?

• Remember. Memory carries across restarts, so projects and preferences don't have to be explained again.
• Learn skills. After a complex task it can turn what worked into a skill, then improve that skill with use. Skills follow the open agentskills.io standard.
• Work from your messaging apps. One gateway connects it to Telegram, Discord, Slack, WhatsApp, Signal and email, as well as the terminal.
• Run scheduled tasks. A built-in scheduler runs jobs you describe in plain English and sends the results to a platform of your choice.
• Split up work. It can start isolated sub-agents for parallel jobs.
• Use tools. More than 40 tools are built in, grouped into toolsets you can switch on and off.
• Run where you want. Seven terminal backends are supported: local, Docker, SSH, Singularity, Modal, Daytona and Vercel Sandbox.
• Use the model you choose. Providers include Nous Portal, OpenRouter, OpenAI and custom endpoints. You switch with hermes model.

How to install Hermes Agent

On Linux, macOS or WSL2:

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

On Windows, in PowerShell:

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

Then reload your shell and start it:

source ~/.bashrc   # or ~/.zshrc
hermes

These are the commands we'd learn first:

hermes setup      # full setup wizard
hermes model      # choose provider and model
hermes tools      # choose which tools are enabled
hermes gateway    # start the messaging gateway
hermes doctor     # diagnose problems
hermes update     # update to the latest version

As with any script piped into a shell, read the installer before you run it.

How we plan to use it

We're building Hermes into Creative Content Engine, our AI video pipeline. We're still on the foundations, so this is the plan, not a finished system.

The rule is that Hermes follows the process we already trust. It reads the project's Bible and workflow before it prepares any job, and it doesn't invent a new process, provider or creative direction.

• Nothing paid runs without approval. Before any paid call, Hermes shows the provider, model, inputs, estimated cost and reason, then waits for a yes.
• Every job is logged. A local job ledger records each attempt, approval, provider response, output and cost. A job can't jump from Draft to Succeeded without a real response from the provider.
• No false reports. Hermes may not say a render, upload or file change happened unless a tool returned a successful result.
• Publishing stays off. Hermes can make content, but it never posts anything. Whether something goes live is always our decision.

We're rolling it out in small steps: audit what the agent can currently do, add the ledger and test it with dummy jobs, build one read-only provider adapter that stops at "awaiting approval", and only then allow a single small paid test.

Our take

Hermes is most interesting as a long-running assistant that can be given real tools, rules and a memory. The more it can do, the more the guard rails matter, so we're starting with the limits before the capabilities.

Frequently asked questions

What is Hermes Agent?

Hermes Agent is an open-source, MIT-licensed AI agent from Nous Research, released in February 2026. It runs on your own machine or server, keeps memory between sessions and learns reusable skills from the tasks it completes.

Is Hermes Agent free?

The software is open source under the MIT licence, so it is free to use. You still pay for whichever model provider or hosting you connect it to.

How do I install Hermes Agent?

Run the one-line installer from the Hermes Agent repository: a curl command on Linux, macOS or WSL2, or a PowerShell command on Windows. Then reload your shell and run hermes to start it.

What can Hermes Agent do?

It remembers across sessions, writes and improves its own skills, runs scheduled tasks, works through Telegram, Discord, Slack, WhatsApp, Signal and email, starts sub-agents for parallel work, and has more than 40 built-in tools.

Which AI models does Hermes Agent work with?

Providers include Nous Portal, OpenRouter, OpenAI and custom endpoints. You switch models with the hermes model command.