Hermes Agent Web UI: Browser Dashboard Setup Guide

The Hermes Agent web UI is a local browser dashboard you open by running hermes dashboard, which starts a web server and loads http://127.0.0.1:9119 on your machine. It is a point-and-click admin surface for the same self-hosted Hermes Agent you would otherwise drive from the terminal or a messaging gateway, and according to the official Nous Research documentation it runs entirely on localhost so no data leaves your computer by default.

What the Hermes Agent web UI is

The Hermes Agent web UI is a browser-based administration panel for managing a single Hermes Agent installation without editing YAML or memorizing CLI flags. Hermes Agent itself is Nous Research's open-source, self-hostable autonomous agent, published at github.com/nousresearch/hermes-agent and described as "the agent that grows with you" — a persistent agent you run on your own machine or server.

People search for it under several names: "hermes agent web ui", "hermes agent webui", and "hermes webui". The official feature is called the web dashboard, and that is the surface this guide covers. There are also community projects with the literal name "Hermes WebUI" — note that those are separate third-party front-ends, not the built-in dashboard documented by Nous Research.

The dashboard is one of three ways to talk to the same agent. You can use the terminal UI (hermes), the messaging gateway (Telegram, Discord, Slack, and others), or the browser dashboard. All three drive the same underlying agent, config, memory, and skills.

How to open the web UI

You open the Hermes Agent web UI by running hermes dashboard from a machine where Hermes is installed. Per the official docs, this "starts a local web server and opens http://127.0.0.1:9119 in your browser," and "the dashboard runs entirely on your machine — no data leaves localhost."

The web dashboard depends on the optional web extra (FastAPI and Uvicorn), and the in-browser Chat tab additionally needs ptyprocess. If you installed Hermes without those extras, add them before launching:

# install the web + pty extras into your Hermes environment
cd ~/.hermes/hermes-agent && uv pip install -e ".[web,pty]"

# launch the dashboard (opens http://127.0.0.1:9119)
hermes dashboard

# bind to a different port, or do not auto-open the browser
hermes dashboard --port 9200 --no-open

The command accepts --port (default 9119), --host (default 127.0.0.1), --no-open, and --insecure. The --insecure flag removes the localhost-only restriction and is flagged as a security risk in the docs, so avoid it unless you fully understand the exposure. If you have not installed Hermes yet, start with our Hermes Agent setup guide first, then come back to the dashboard.

Pairing and authentication

On localhost the Hermes Agent web UI has no login, but binding it to a non-loopback address automatically engages an authentication gate. The docs state plainly that "binding the dashboard to a non-loopback address engages its auth gate" — so a default 127.0.0.1 session is open on your own machine, while a dashboard reachable over the network requires sign-in.

When auth is engaged, the documentation lists three providers: Nous Portal OAuth (recommended for any internet-facing dashboard), username/password (for trusted networks only), and self-hosted OIDC (for custom identity providers). For a backend reachable beyond your own machine — a VPS or public host — Nous Research recommends the OAuth path described in its Run Hermes with Nous Portal guide.

There are two distinct meanings of "pairing" here, and it helps to keep them separate:

• Admin pairing (dashboard access): for the desktop client connecting to a remote backend, you protect the remote dashboard with an auth provider and sign in once. After sign-in, the session is reused for the chat WebSocket automatically — per the docs, "there is no token to copy or paste."
• Channel pairing (messaging users): the dashboard's Pairing/Channels page lets you approve and revoke messaging users — for example onboarding a Telegram or Discord user to your gateway — with full parity to the hermes pairing command, all from the browser.

If your main goal is messaging, our Hermes Agent Telegram setup guide walks through connecting that channel end to end, and the dashboard can manage the same approval flow visually.

What you can do in the dashboard

The Hermes Agent web UI started as a session viewer and grew into a full administration surface that replaces most config-file editing. According to the official feature docs, the core navigation follows the same pattern across pages: "Config, API Keys, Skills, MCP, Models, and the Chat tab all follow it," alongside Sessions, Logs, Analytics, Cron jobs, Profiles, Webhooks, Pairing, Channels, and System administration.

In practice, the dashboard lets you do the day-to-day operations work without SSHing in to edit config.yaml:

• Chat tab: talk to the agent directly in the browser (this is the page that needs the pty extra).
• Config and Models: change settings and switch your LLM provider/model without touching files. For picking a model, see our best models for Hermes Agent guide.
• API Keys and Credentials: add, rotate, and manage provider keys from one screen.
• Skills: view and toggle the skills your agent can use. Learn how those work in our Hermes Agent skills guide.
• MCP: an admin catalog with enable/disable toggles for Model Context Protocol servers.
• Channels and Pairing: wire up and approve messaging platforms point-and-click.
• Cron, Webhooks, Sessions, Logs, Analytics: schedule jobs, create hooks, and monitor what the agent is doing.

Everything you configure here lives in the same self-hosted instance you would manage from the terminal. If you are running Hermes on your own box or a VPS, our self-hosted Hermes Agent guide covers the deployment side that the dashboard then sits on top of.

Web UI vs Telegram vs CLI

The web UI, the Telegram/gateway interface, and the CLI are three front-ends to one agent, each better at different jobs. The CLI is the fastest way to a running agent and the richest interactive experience; Telegram and the other gateway channels are best for talking to the agent from your phone; the web UI is best for configuration, monitoring, and visual administration.

Most operators use more than one: the docs note Hermes "has two entry points — start the terminal UI with hermes, or run the gateway and talk to it from Telegram, Discord, Slack, WhatsApp, Signal, or Email" — and the dashboard rounds that out with a visual admin layer. For a deeper architectural comparison with the OpenClaw lineage, see our OpenClaw vs Hermes Agent breakdown.

Limitations and Tradeoffs

The Hermes Agent web UI is a single-instance control panel, not a hosted service or a multi-tenant console. It manages one agent on the machine where you run hermes dashboard; it does not spin up servers, host the agent for you, or manage a fleet of remote agents from one place.

Two tradeoffs are worth calling out. First, exposing the dashboard beyond localhost meaningfully changes your security posture — the --insecure flag and non-loopback binds open it up, and you should only do that behind a properly configured auth provider (ideally OAuth) and network controls. Second, the dashboard's usefulness depends on the optional extras being installed; without the web extra it will not start, and without the pty extra the Chat tab will not work.

When NOT to lean on the web UI: if you only need quick one-off interactions, the CLI is faster; if you only chat from your phone, the gateway is simpler. And feature details evolve quickly in this fast-moving open-source project, so verify exact flags and pages against the official docs for your installed version rather than trusting any single blog (including this one). Remote OpenClaw does not host or deploy Hermes for you — we sell the personas, skills, and bundles you load into it. Browse those on the marketplace.

Related Guides

• Hermes Agent Setup Guide
• Self-Hosted Hermes Agent Guide
• Hermes Agent Telegram Setup
• OpenClaw vs Hermes Agent