Every Hermes Dashboard Command I Use (Tested)

Q: How does Hermes remember what I worked on?

My conversations are linked to my Obsidian vault, which I call the memory galaxy. Every session and build is logged there, so memory commands read real logs rather than guessing.

Every Hermes dashboard command in my system does one of six jobs: it builds something, it remembers something, it shows me something, it opens something, it lists my agents, or it makes the voice talk.

This is the reference version of that list — every command I actually run, what it returns, and which part of the stack does the work.

No theory, just the commands.

The six command types at a glance

Command type	Example	What comes back
Build	"Build me a galaxy I can swirl with my mouse"	A working, interactive file in the builds panel
Memory	"What have I been working on this week?"	A read of your real activity logs
Metrics	"Show me my numbers"	The numbers panel on screen
Agents	"Show me my team"	Every agent configured in your system
Browser	"Open up juliangoldie.com"	The site opens on your machine
Archive	"Show me everything I've built"	The full builds gallery

Everything below is one of those six, phrased differently.

🔥 Want the exact setup I used to get these results? Inside the AI Profit Boardroom, I've got a full Hermes Jarvis and Agent OS section with step-by-step video tutorials, the install zip and the prompts. Plus weekly coaching calls + 2,800+ members building real automations. → Get access here

Build commands

This is the category that makes people sit up.

"Jarvis, build me a galaxy I can swirl with my mouse."

It comes back with "Built and running, sir" and the galaxy is there, in the browser, spinning under your cursor.

The important detail is the interaction clause.

"Build me a galaxy" gets you a picture of one, and "build me a galaxy I can swirl with my mouse" gets you something you want to play with.

Always say what you want to do with the thing, not just what the thing is.

Things I've built with a single command: a habit tracker I still use, a full website, a Japanese flashcard game, a meditation timer and videos.

I don't code.

Every one of those came out of a command and got filed into the builds section automatically.

The builds panel is the underrated half of this.

Anything you create is still there later, so a command isn't a one-off demo — it's an asset you can reopen.

I go deeper on what's worth building in my Hermes agent use cases breakdown.

Memory commands

"Jarvis, what have I been working on this week?"

Mine answered: "you've been tinkering with thumbnail work this week, sir — the thumbnails log shows recent edits there."

That's not a guess, and that's the whole point.

My conversations with Hermes and Hermes Jarvis are linked to my Obsidian memory, which I call the memory galaxy.

Everything I do gets written to a log I can go back to later.

So memory commands read a file, not a vibe.

If your memory commands feel hallucinated, your memory link isn't wired up — fix that before blaming the model.

The Hermes second brain setup covers how to connect the vault.

Dashboard view commands

"Jarvis, show me my numbers."

"Pulling up the numbers, sir" — and the metrics view appears.

"Jarvis, show me my team."

This pulls up every agent ready inside the system, so you see Claude, OpenClaw and Hermes all together in one panel.

It's a genuinely useful sanity check when you're not sure what's actually configured versus what you meant to configure.

"Show me everything I've built."

"Here's everything we've built, sir" — and the gallery opens.

You can click straight into any previous build from there.

Browser and computer use commands

"Jarvis, open up juliangoldie.com."

"Opening Julian Goldie's site now, sir" — and the website opens.

That works because computer use and browser use are built into the system, which is powerful in its own right.

It also means the same capability is available inside longer tasks, not just as a party trick.

A build command can go and look at a real page before it builds something based on it.

Voice and mode commands

These aren't spoken commands so much as dashboard controls, but they change how every other command behaves.

You can switch between voices from the dashboard, so the assistant sounds however you want it to sound.

You can switch between auto and agent mode.

Agent mode is what you want for build commands, because the agent needs multiple steps and tool access.

Auto mode is fine for quick lookups.

You've got a wake word, and I use both "Jarvis" and "Hermes."

Say it and it switches on, say it again and it switches off.

That gives you wall mode — the assistant sits in the background ready and waiting, and you type into it whenever you'd rather not talk.

"Jarvis, tell me a joke that's about AI."

Mine said the neural network broke up with the database because she had too many issues and wouldn't commit to a schema.

It's a throwaway command, but it's my standard check that the ElevenLabs voice layer is alive before a real session.

The panels your commands feed into

A command is only half the system.

Here's what the dashboard does with the results.

Panel	What lives there
History	Every conversation you've had, in full
Builds	Every website, game, tool and video created
Model controls	Which model answers, swappable from the dashboard
MCPs	Links Hermes to external tools
Control room	The wider system view
Goal mode	Agent chases an outcome instead of one task
Chat mode	Plain typed conversation
Mission control	Agent teams and Paperclip
Agent group chat	Hermes talking to Claude and Gemini

Mission control and the group chat are where this stops being a voice toy.

Hermes talking to Claude and Gemini in one thread is a different kind of leverage — I break that down in Hermes agent mission control.

🔥 Want my full Hermes dashboard command setup? The complete Agent OS — dashboard, Jarvis voice layer, wake word config, install zip and prompts — is inside the AI Profit Boardroom, updated daily with new tutorials. Ask questions and get help in real time from me and 2,800+ members. → Get the Agent OS here

What the stack behind the commands is

Three parts, and that's it.

Hermes agent is the open source project, and all of its agentic capabilities are built into this.

ElevenLabs generates the voice response.

Claude runs in the background handling the improvements and the system I've built around it.

If you want to build your own version, that's genuinely the recipe: combine Hermes with ElevenLabs and get Claude to help you in the background.

I've done something similar with free Claude Code too, using an agent factory that builds things live while you speak or type at it.

That one runs Qwen 2.5 Coder with Ollama, so it's a free local model doing the work — which is mind-blowing when you sit with it for a second.

Local models keep getting better, and that changes the maths on all of this.

Command troubleshooting

The command answered instead of building. You're in auto mode, not agent mode.

The memory command invented an answer. Your Obsidian log link isn't syncing.

The voice didn't respond. Check the ElevenLabs key before touching anything else — run the joke command as a test.

The wake word keeps triggering. Switch it off when recording; wall mode is optional, not mandatory.

The build is ugly. Add the interaction and the style to the command itself rather than fixing it afterwards.

And the honest one: sometimes it just isn't that smooth.

I improve this thing every single day and it still misfires occasionally.

For heads-down production work I'd reach for Paperclip or a structured agent team, because you'll create more, faster.

As a fun party trick with AI that quietly turns into real builds, this is my favourite bit.

FAQ

What is a Hermes dashboard command?

A Hermes dashboard command is a spoken or typed instruction given through the Hermes dashboard. The agent runs it using its agentic tools and files any result — a website, game, video or report — into the dashboard's builds section.

What is the wake word for Hermes dashboard commands?

I use two: Jarvis and Hermes. Saying either switches the listener on, and saying it again switches it off so the dashboard can sit quietly in wall mode.

Can Hermes open websites from a dashboard command?

Yes. Computer use and browser use are built into the system, so a command like "open up juliangoldie.com" actually opens the site on your machine.

How does Hermes remember what I worked on?

My conversations are linked to my Obsidian vault, the memory galaxy. Every session and build is logged there, so memory commands read real logs rather than guessing.

Which models can I run behind the dashboard?

You manage the model from the dashboard itself. My team panel shows Claude, OpenClaw and Hermes side by side, and I've run local models like Qwen 2.5 Coder through Ollama for a similar free setup.

Does every Hermes dashboard command work first time?

No, and I'd rather be honest about it. Sometimes it isn't smooth and a command misfires. It's a system I improve daily rather than a finished product.

About Julian

I'm Julian Goldie — AI entrepreneur, SEO expert, and founder of the AI Profit Boardroom (2,800+ members). I help business owners scale with AI agents, automation, and SEO.