agentos.guide › blog
Hermes3D is a free, open-source 3D virtual office for AI agents that you run on your own machine, and the repo is github.com/iamlukethedev/Hermes3D.
It is MIT licensed, and it is an unofficial community project built by LukeTheDev, so it is not made by Nous Research.
The quick start is git clone https://github.com/iamlukethedev/Hermes3D hermes3d, cd hermes3d, npm install, cp .env.example .env and npm run dev, and then you open http://localhost:3000.
I'm Julian Goldie, and I have not tested Hermes3D myself yet.
This is a docs-based guide built from the README, the tutorial and the docs folder, which I read on 14 October 2026.
The video below is my free Agent OS course, because I have not made a Hermes3D video, and this page shows where a 3D office fits in that kind of setup.
Here is the whole thing before any explanation.
npm install..env.example to .env.http://localhost:3000, connect to ws://localhost:18789 and choose the Demo backend.The rest of this page explains each step and shows where a 3D office sits in an Agent OS.
Hermes3D is a screen that turns your AI agents into workers in an office you can walk around.
The README describes it as a "3D virtual office for AI agents" that runs on your own infrastructure.
Most agent tools show you logs and terminal output.
Hermes3D shows you desks, rooms and characters, so you can see who is working and who is waiting.
The README says the gateway is only the control plane, and the office is the product.
The repo's vision file says Hermes provides the intelligence and Hermes3D provides the visual layer.
These are the facts from the repo and its README on 14 October 2026.
| Item | Detail |
|---|---|
| Official repo | github.com/iamlukethedev/Hermes3D |
| What it is | It is a 3D virtual office for AI agents that you host yourself. |
| Maker | It is built by LukeTheDev, an independent developer. |
| Official status | It is unofficial, and it is not made by Nous Research. |
| Price and licence | It is free and open source under the MIT licence. |
| Language | It is written in TypeScript. |
| Created | The repo was created on 17 August 2026. |
| Releases | There is one release, v1.0.0, dated 20 August 2026. |
| Last push | The last push I could see was on 29 August 2026. |
| Stars and forks | It had 570+ stars and 120+ forks when I checked. |
| Requirements | Node.js 20 or newer and npm 10 or newer are recommended. |
One line on the GitHub page can send you the wrong way.
The repo's short "About" text calls Hermes3D an open-source 3D engine for games and simulations.
I think that line is misleading.
The README, the tutorial and the vision file all describe a 3D office for watching and steering AI agents.
Nothing in those files is about building games.
The README has a clear notice about this.
It calls Hermes3D an independent, community-driven frontend.
It says the project is not affiliated with, endorsed by or maintained by the teams behind the agent backends it connects to.
In plain words, Nous Research does not make it and does not support it.
It is built and maintained by LukeTheDev, who also makes Herald OS.
An Agent OS is a dashboard that runs a crew of AI agents with shared memory.
I think of it as five layers, and Hermes3D only touches the top two.
| Layer of an Agent OS | What does that job | Where Hermes3D sits |
|---|---|---|
| The brain | The AI model your agents think with. | Hermes3D has no model, so it adds nothing here. |
| The agents | Hermes, Claude and the other agents that do the work. | Hermes3D has no agent of its own, so it needs one of yours. |
| The memory | The shared notes and history your agents read. | Hermes3D keeps runtime state in the backend and only stores its own screen settings. |
| The cockpit | The dashboard where you give orders and approve actions. | The docs say Hermes3D covers chat, approvals and agent settings. |
| The window | The view that shows you what every agent is doing. | This is the job Hermes3D is built for. |
Hermes3D will not make a weak agent smarter.
It will not give you an agent if you do not have one.
What it can do, according to the docs, is make a working crew visible.
If you have not built an Agent OS yet, my Agent OS build guide shows the whole stack.
That video is my walkthrough of Hermes Pixel Office, which is a different visual office, and I have not made a Hermes3D video.
🔥 Want the exact Hermes agent 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 README recommends Node.js 20 or newer and npm 10 or newer.
You need git for the clone.
You also need a backend, because the README says Hermes3D does not install or build Hermes for you.
| Backend | How it connects | When I would pick it |
|---|---|---|
| Demo gateway | A built-in mock gateway with demo agents on ws://localhost:18789. | I would start here, because it needs no agent. |
| Hermes through the adapter | The bundled adapter calls the Hermes HTTP API at http://localhost:8642. | I would use this when Hermes runs on the same machine. |
| Hermes Agent (direct) | Studio talks to a hermes serve backend with no adapter process. | I would use this for a remote Hermes host over Tailscale. |
| Custom runtime | Studio proxies a direct HTTP runtime through its own route. | I would leave this to developers with their own orchestrator. |
If Hermes is not installed yet, my Hermes crash course gets you there first.
These are the README's commands with the repo URL filled in.
git clone https://github.com/iamlukethedev/Hermes3D hermes3d
cd hermes3d
npm install
cp .env.example .env
The last line gives you your own settings file to edit.
I would always run the demo before I connect a real agent.
You run these two commands in two separate terminal windows.
npm run demo-gateway
npm run dev
You open http://localhost:3000 in your browser.
You connect Studio to ws://localhost:18789 and choose the Demo backend.
The README says the demo starts a mock gateway with demo agents, streaming chat, session previews and office presence.
Studio is the name the docs use for the web app and its small server.
If the 3D view is heavy on your machine, the README says you can switch to the 2D pixel office in Settings under Office renderer.
The default path uses the bundled adapter.
The tutorial draws the chain as browser, then Studio, then the adapter, then the Hermes HTTP API.
The adapter expects the Hermes HTTP API at http://localhost:8642 by default.
The official Hermes docs say that API server is off until you enable it and give it a key, and then you start the Hermes gateway.
You then set the Hermes values in the Hermes3D settings file.
HERMES_API_URL=http://localhost:8642
HERMES_API_KEY=
HERMES_ADAPTER_PORT=18789
You put your own key in that file, and you never share or commit it.
You run the adapter and Studio in two terminals.
npm run hermes-adapter
npm run dev
You connect to ws://localhost:18789 and choose the Hermes backend.
The docs say the demo gateway and the adapter share port 18789, so you only run one of them at a time.
There is also a direct path with no adapter, and the direct-path guide covers it for a remote Hermes host.
I have not run either path with Hermes3D, so follow the repo's own guides line by line.
The tutorial calls this the most common wrong turn.
Port 9119 is the port the Hermes dashboard prints when it starts.
With the adapter path, Hermes3D cannot connect to it, because the tutorial says that port speaks a different protocol.
You point Hermes3D at the adapter on port 18789 instead.
The repo has a diagnostics command, npm run doctor, and the tutorial says it flags this mistake.
The README lists these features, and I have not tested them.
| Feature | What the README says it does |
|---|---|
| 3D office | You watch agents work in real time in a shared office with desks, rooms and animations. |
| 2D pixel office | You get a Gather-style pixel view of the same agents for low-power machines. |
| Fleet and chat | You manage agents and chat with them, with updates streamed from the gateway. |
| Approvals | You handle approvals in the same app as chat and settings. |
| Standups | You run standups with agents connected to GitHub and Jira. |
| Pull requests | You review pull requests from inside the office. |
| QA and logs | You monitor QA pipelines and logs without leaving the workspace. |
| The gym | You train agents in the gym to develop new skills. |
| The janitor | You reset sessions and clean context with the janitor system. |
| Office builder | You edit and publish office layouts at /office/builder. |
| Voice replies | You can add your own ElevenLabs keys to enable voice replies. |
| SOUNDHERMES | You get a Spotify jukebox inside the office. |
The repo's code also contains a kanban screen, although the README does not document it.
There is an optional bridge plugin that makes the office react to chats you have in the Hermes desktop app, the TUI or the CLI.
The docs say that plugin only observes, so it never changes or delays a conversation.
🔥 Want the exact Hermes agent 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.
This is the part that matters most for an Agent OS.
The adapter docs say the main Hermes agent can act as an orchestrator with these tools.
| Tool | What the adapter docs say it does |
|---|---|
spawn_agent | It creates a specialist sub-agent. |
delegate_task | It sends work to a specific agent. |
list_team | It lists active agents with their names and roles. |
configure_agent | It updates an agent's name, role, instructions or settings. |
dismiss_agent | It removes an agent from the team. |
read_agent_context | It reads another agent's recent conversation for coordination. |
The docs say sub-agents appear in the office as separate characters and keep their own conversation state.
That is the same idea I use in my own Agent OS, where one agent hands jobs to others.
The difference is that Hermes3D shows the handover as people in a room.
That video is my test of Herald OS, which is another Hermes project by the same developer.
This is a plan, and it is not a test I have run.
I would judge it on one question, which is whether I spot a stuck agent faster than I do today.
The docs spend real time on this, and I am glad they do.
| Setup | What the docs say to do |
|---|---|
| Gateway on another machine | You publish the gateway with tailscale serve and use the wss:// address in Studio. |
| SSH instead of Tailscale | You forward port 18789 over SSH and keep using ws://localhost:18789. |
| Studio on your network | You set STUDIO_ACCESS_TOKEN before Studio is reachable beyond localhost. |
| Production | You set UPSTREAM_ALLOWLIST to restrict the gateway hosts Studio may proxy to. |
The README says the gateway URL and token are loaded into browser memory at runtime.
It tells you to treat the browser as part of your trust boundary.
The tutorial prefers Tailscale Serve or an SSH tunnel over exposing gateway ports to the internet.
The full walkthrough for two machines is in the repo's tutorial.
The README lists its own limits.
I would add two more.
The project has one release and no pushes since late August 2026, so check the repo for activity before you rely on it.
I have not tested it, so I cannot tell you how stable it is.
I have used the other three tools in this table on my own machine.
| Tool | What it is | My hands-on status |
|---|---|---|
| Hermes3D | A 3D office with a 2D mode, by LukeTheDev. | I have read the docs, and I have not tested it. |
| Hermes Pixel Office | A pixel office plugin by Teknium. | I have used it and made a video on it. |
| Herald OS | An OS-style Hermes front-end from the same developer as Hermes3D. | I have used it, and it is an alpha. |
| My Agent OS | My own dashboard for Hermes, Claude and other agents under one memory. | I run it daily, and it is my own product. |
My Agent OS is available inside the AI Profit Boardroom, and it is not on public GitHub.
It is my own product, so weigh my view of it with that in mind.
My ranked list of interfaces is in the best Hermes Agent UI guide.
The sibling guides from this batch are Hermes Agent dashboards and the best UI for Hermes Agent.
You should try it if you already run Hermes and you want to see your agents instead of reading logs.
You should run the demo if you are curious, because it needs no agent and no account.
You should wait if you are new to agents, because getting one agent working comes first.
You should also wait if you need a tool with official support or regular releases.
Hermes3D is a free, open-source 3D virtual office for AI agents that you run on your own infrastructure. Agents appear as workers in an office, and the docs say you can chat with them, handle approvals and watch their activity. It is MIT licensed and built by LukeTheDev.
Clone github.com/iamlukethedev/Hermes3D, then run npm install, cp .env.example .env and npm run dev, and open http://localhost:3000. For the demo, also run npm run demo-gateway, connect to ws://localhost:18789 and choose the Demo backend. The README recommends Node.js 20 or newer and npm 10 or newer.
No. The README calls it an unofficial, independent, community-driven frontend that is not affiliated with or endorsed by the teams behind the backends it connects to.
No. The repo's short About line calls it a 3D engine for games and simulations, but the README, the tutorial and the vision file all describe a 3D office for AI agents.
It sits at the top as the window and part of the cockpit. It has no model and no agent of its own, so it needs a backend such as Hermes. The docs say it covers chat, approvals and agent settings on top of the visual office.
No. This guide is based on the repo's README, tutorial and docs, read on 14 October 2026. Julian has used Hermes Pixel Office, Herald OS and his own Agent OS.
The repo has one release, v1.0.0 from 20 August 2026, and the last push visible on 14 October 2026 was on 29 August 2026. Check the repo for recent activity before you rely on it.
Get one agent working, run the demo, and check the repo for activity before you depend on it, and you will have an honest first look at Hermes3D.
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.
→ Herald OS: The Free Hermes Agent OS From The Same Developer
→ The Best Hermes Agent UI, Ranked
→ How To Build Your Own AI Agent OS
→ Hermes AI Agent: The Full Crash Course
→ Hermes Agent OS: The Free Options
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 👉