agentos.guide › blog
Hermes Herald AI Agent: A Five-Minute Quick-Start And How It Fits Your Agent OS
The Hermes Herald AI agent is Hermes Agent running as the whole interface of Herald OS, a free, open-source, agent-native operating system by Luke The Dev.
I had Claude set it up for me in about two minutes, and my existing Hermes automations showed up in it straight away.
In this quick-start I will show you the five setup steps, how Herald talks to Hermes, every page the agent drives and where it fits in an Agent OS.
I will also clear up the naming clash with the Hermes "Herald Release" in one short section.
What the Hermes Herald AI agent is in one minute
Herald OS calls itself "an agent-native operating system, with Hermes Agent as the interface".
You talk or type to Hermes, and it works across the whole computer for you.
It opens apps, finds and organises files, watches what is running, remembers what matters, runs routines and builds software while you watch.
It is built by Luke The Dev, an independent developer, and it is not affiliated with Nous Research.
It is free and open source under the MIT licence.
It is alpha, version 0.1, and the first release landed on 6 October 2026.
I used to run Hermes in the terminal back in March and April, and it was messy.
Herald OS is the new way, with a clean interface you can personalise, and I think it looks nicer than Hermes Desktop.
The five-minute Hermes Herald AI agent quick-start
You need an Apple Silicon Mac on macOS 13 or later for this route.
| Step | What you do | Time it took me |
|---|---|---|
| 1 | Install Hermes Agent with the official installer and run hermes setup to pick a model provider. | A few minutes if Hermes is new to you, and zero if you already run it. |
| 2 | Download the Herald OS DMG from the GitHub releases page and drag it into Applications. | About a minute. |
| 3 | Run the xattr command from the README, because the Mac build is not notarized by Apple yet. | A few seconds. |
| 4 | Open Herald OS, which takes over the screen, and use Cmd+Ctrl+F to toggle fullscreen and Cmd+Q to quit. | A few seconds. |
| 5 | Ask Hermes a question, then ask it to move a file and check that an approval card appears. | About a minute. |
For the model, Hermes can use a Nous Portal account or an API key for OpenRouter, OpenAI, Anthropic, a local model and others.
If you build Herald from source, the README asks for Hermes Agent 0.21.3 or later, so run hermes update first.
I did not type any of this myself, because I asked Claude to set it up and it was done in about two minutes.
If Hermes answers with shell commands instead of its system tools, run hermes plugins list and check that herald-os-bridge is enabled.
There are Linux routes too, including a full Herald OS install for a spare PC, an Arch package, Omarchy support, a tarball and a virtual machine.
How Herald talks to Hermes, and why your Agent OS keeps working
This is the part that matters most if you already run an Agent OS.
Herald OS does not fork Hermes and does not rebuild the agent in its own code.
It starts your normal Hermes install in the background with hermes serve, listening on your own machine only.
| Layer | Who owns it | What that means for your Agent OS |
|---|---|---|
| The machine | The Herald OS shell | It handles the window, fullscreen, the file system, app discovery, system stats and notifications. |
| The experience | The Herald OS interface | It draws the pages, the dock, the command bar and the approval cards. |
| The work | Your Hermes install | It keeps your sessions, model calls, tools, skills, memory, scheduled jobs and approvals. |
| The bridge | The herald-os-bridge plugin | It gives Hermes typed system tools with permission tiers and an audit log. |
The bridge plugin is the piece that gives the agent hands.
On a Mac it links itself into your Hermes plugins folder on first start and switches itself on.
It registers a small toolset for system info, processes, disk usage, file search, apps, network, logs, opening things, moving files and stopping programs.
It also adds a tool that lets Hermes drive the Herald interface itself, so it can open pages, add memories, run automations and start missions.
Because Hermes owns the work, everything else in your Agent OS keeps seeing the same sessions, memory and scheduled jobs.
🔥 Want the exact running Hermes inside Herald OS setup I run inside my Agent OS?
Inside the AI Profit Boardroom you get the installable Agent OS, step-by-step video tutorials, four coaching calls a week and 3,400+ members building real automations.
Automations sync because it is the same Hermes underneath
This was the moment that sold me.
My automations synced across Herald OS, Hermes Desktop and my Agent OS straight away.
The reason is simple, because the Automations page reads and writes Hermes's own scheduled jobs.
There is no second automation system to keep in step.
The ones that showed up on my machine were a daily 6am to-do recap, a daily Japanese learning routine and morning inbox summaries.
You can manage each one and switch it on or off from a single page.
The docs also describe automations that run when you log in, come back after a break, hit low battery or see a program crash.
So in an Agent OS workflow, Herald OS becomes one more front end on the same brain rather than a new silo.
Every page the agent drives
Here is the quick tour, with the agent side first.
- The Overview greets you with "Good morning, Julian. Your day already in motion." and shows today's schedule from your scheduled tasks.
- The Overview also shows what needs attention, recent desktop screenshots and a voice input button.
- "Pick up where you left off" suggests up to three threads from the names of your recent documents, projects, chats and calendar events.
- That feature is off until you switch it on, and it reads names and dates only, never file contents.
- The Hermes page is normal Hermes chat with Preview and Sources, so you can add files Hermes reads.
- Spaces split your work, and my screen showed Ideas, Work and Personal, with room to add one per client.
- Missions take an active goal like "build out a new website" and plan it step by step.
- Memory shows everything Hermes remembers, and it showed my Telegram details on camera, so check it before you screen-share.
- Files lists recent, favourites, projects, downloads, shared and deleted items.
- Automations manages your scheduled routines, which sync with every other Hermes front end.
- Connections is meant for tools and apps, but it did not work or sync for me in this alpha.
The sidebar is resizable, and the interface was cut off slightly at the top on my screen.
Approvals: how the agent asks before it acts
Every system tool sits in one of four tiers.
| Tier | What happens | Examples |
|---|---|---|
| Read | It runs straight away and gets logged. | System info, processes, disk usage and file search. |
| Act | It runs straight away, gets logged and shows a notification. | Opening an app or a project folder. |
| Mutate | It asks first, and you can allow it for the session or always. | Creating folders, moving and renaming files. |
| Destructive | It asks every single time. | Stopping a process or moving files to the Trash. |
Secret locations like SSH keys, keychains and Hermes's own credentials are always refused.
Herald OS moves files to the Trash and never deletes permanently.
Every action is written to an audit log that you can read under Settings and Privacy.
In an Agent OS, that log is gold, because you can see exactly what a background job touched overnight.
🔥 Want the exact running Hermes inside Herald OS setup I run inside my Agent OS?
Inside the AI Profit Boardroom you get the installable Agent OS, step-by-step video tutorials, four coaching calls a week and 3,400+ members building real automations.
The settings that shape your agent
| Setting | What it controls |
|---|---|
| Autonomy | You pick Ask first, Balanced or Independent, which sets how often Hermes checks in. |
| Preferred model | It picks the model Hermes uses for most tasks, and mine showed GPT-6 Astra through a CLI I had plugged in. |
| Run on | It sets where agents run tasks, such as this device or a runtime like Docker. |
| Background agents | It sets how many background agents run at once, from 1 to 32. |
| Remember preferences | It lets Hermes remember your settings and learn from your choices. |
| Ask before sending messages | It makes Hermes confirm before it messages anyone for you. |
| Activity history | It shows background work you may have forgotten about. |
| Usage | It shows tokens used, your plan and your token balance. |
| Voice | It switches engine and speed, tests the voice and lists every voice command. |
| Software and Plugins | It adds coding agents and local models like Claude Code, Codex and LM Studio, and plugins from a GitHub link. |
The General section covers launch fullscreen, default Space, default and projects folders, do not disturb, stay awake, screensaver, dictation and quit session.
There is a bell notification sound, and do not disturb keeps things quiet.
There are colours and themes, plus a Hermes Agent section, privacy, notifications, networks, storage and accessibility.
Voice: the Jarvis layer
On a Mac you press Alt+Space to talk, or you can switch on a "hey Hermes" wake word.
Hermes answers out loud and keeps listening for a follow-up.
Simple commands like "open missions" run locally without spending tokens.
The free engine is the default, and the docs put a turn at about two seconds.
The opt-in Live engine uses OpenAI's gpt-live-1, which the docs price at $0.05 a minute with a daily cap.
You can switch between local voice and live voice, set the speed, test it and read the voice command list.
Herald OS vs the Hermes "Herald Release"
People mix these two up all the time, so here is the clean split.
Nous Research codenamed Hermes Agent v0.20.0 "The Herald Release", and it shipped on 3 August 2026.
That update added conversational voice with barge-in, A2A agent-to-agent support, signed outbound webhooks and grounded citations.
I tested it inside my Agent OS in The Herald Upgrade guide.
Herald OS is a separate, independent project by Luke The Dev, and it is not affiliated with Nous Research.
It runs on top of Hermes, and its from-source route asks for Hermes 0.21.3 or later, which is newer than the Herald Release.
Where it fits in my Agent OS workflow
My Agent OS already runs Hermes, Claude Code and a stack of automations.
Herald OS does not replace any of that.
It gives me a beautiful full-screen front end on the same Hermes, with the same memory, models and scheduled jobs.
I would use it as the daily driver for chat, missions and voice, while the Agent OS keeps the heavier pipelines.
If you want Hermes on a free brain first, my Hermes free forever guide shows you how.
For the full tour of the operating system itself, read my main Herald OS explainer.
Tips and honest limits
It is alpha, so keep backups of anything you let an agent touch, which the README says too.
Connections did not work for me, although the code is open and you could ask Claude to fix it.
Herald OS is free, but the model may cost money, so check the Usage page during your first week.
Start on Ask first, and only loosen autonomy once you trust what the agent does.
If you are happy with Hermes Desktop, you do not have to switch, because your automations will show up in both.
Also On Our Network
FAQ
What is the Hermes Herald AI agent?
It is Hermes Agent running as the interface of Herald OS, a free, open-source, agent-native operating system by independent developer Luke The Dev. Herald OS runs your normal Hermes install and adds system tools through a plugin called herald-os-bridge.
Is Herald OS made by Nous Research?
No. Herald OS is an independent project and is not affiliated with Nous Research. Nous Research makes Hermes Agent and codenamed Hermes v0.20.0 the Herald Release in August 2026, which is a separate thing.
Do my Agent OS automations work in Herald OS?
Yes. Herald OS reads and writes Hermes's own scheduled jobs, so my automations synced across Herald OS, Hermes Desktop and my Agent OS straight away.
Which model does the Hermes Herald AI agent use?
Whatever your Hermes is set up with, such as a Nous Portal account or an API key for OpenRouter, OpenAI, Anthropic or a local model. You can change it in Settings or with hermes model.
Is Herald OS safe to run?
Changes go through permission tiers, trashing files and stopping programs ask every time, and secret folders are always refused. It is still alpha, so keep backups and start on Ask first.
Final word on the Hermes Herald AI agent
Install Hermes, add Herald OS, keep approvals on and let your existing automations follow you in.
Run it beside your Agent OS for a week, and you will know whether the Hermes Herald AI agent earns a permanent place on your screen.
About Julian
I'm Julian Goldie, an AI entrepreneur, SEO expert and the founder of the AI Profit Boardroom, which has 3,400+ members.
I help business owners scale with AI agents, automation and SEO.
My YouTube channel has 400,000+ subscribers, and I run Goldie Agency, a seven-figure SEO agency.
Related guides on agentos.guide
→ The Herald Upgrade: Hermes v0.20
Get the whole Agent OS
The installable system behind every guide on this site — dashboard, agents, memory and pipelines — updated as new tools land, with 3,400+ members and four coaching calls a week.
Join the AI Profit Boardroom →📺 Video notes + links to the tools 👉
🎥 Learn how I make these videos 👉