Hermes Agent Workspace and Dashboard

Workspace or Dashboard? (Terminology)

Community discussion uses both terms interchangeably. You will see “Hermes workspace” and “Hermes dashboard” used to mean the same thing: the browser-based management UI for a running Hermes Agent. The upstream Hermes documentation is the authoritative source for whatever the current official name is; this guide uses both terms to match what people search for.

By default the workspace is available at http://127.0.0.1:9119. It binds to localhost only, so it is not reachable from other machines without additional setup.

What the Workspace Lets You Do

The workspace is focused on configuration and monitoring, not interactive chat. It surfaces the following tabs:

• Status — live health of the gateway process and connected platforms
• Config — review and edit the agent’s live configuration
• API Keys — manage API keys used by the agent
• Sessions — read-only browser of active and past sessions (no send-input from this view)
• Logs — recent model calls, tool invocations, and errors
• Analytics — usage statistics for the running agent
• Cron — scheduled tasks the agent runs automatically
• Skills — manage which skills are enabled for the agent

The workspace also surfaces the shell-exec approvals queue when approvals.mode is set to “manual” or “smart”. Any shell command, file edit, or network action the agent wants to run that requires human sign-off appears here for you to approve or reject.

What the workspace does not include: a send-message chat pane (sessions is read-only). To actively converse with the agent, use a connected channel — Telegram, Discord, WhatsApp — or the OpenAI-compatible API endpoint described below.

How to Launch the Dashboard

The dashboard is a separate CLI command — it is not part of the hermes gateway process. You start it independently:

1. Install the web extra. The dashboard requires an optional dependency set. If you installed Hermes without it, run:
   pip install hermes-agent[web]
   Without this extra, hermes dashboard will not start.
2. Start the dashboard. From any terminal on the host running Hermes:
   hermes dashboard
   This starts the UI server on port 9119 and opens a browser tab at http://127.0.0.1:9119.

You can run hermes dashboard and hermes gateway side by side on the same host — they are two separate processes, not a single combined process. The dashboard connects to the running gateway to read status and config.

The dashboard has no built-in authentication. It relies on the localhost binding for security: because it binds to 127.0.0.1 only, it is not reachable from outside the machine unless you explicitly expose it. Do not expose port 9119 unauthenticated on the public internet.

To access the dashboard remotely, use an SSH tunnel (recommended) or a reverse proxy with authentication:

• SSH tunnel (recommended for personal use) — run ssh -L 9119:127.0.0.1:9119 user@your-server from your local machine, then open http://127.0.0.1:9119 in your browser. The tunnel forwards the port securely without any firewall changes or public exposure.
• Reverse proxy with auth — Caddy or Nginx terminating HTTPS and forwarding to port 9119, with HTTP Basic Auth or similar access control. Never put a proxy in front of port 9119 without adding authentication.

Accessing the Workspace

If you run Hermes Agent yourself (Docker or bare metal), the workspace is launched with hermes dashboard and is at http://127.0.0.1:9119 by default. It is only accessible from the machine running the command unless you use a tunnel or reverse proxy.

For the full step-by-step access setup, see the Hermes Agent web UI guide, which covers reverse-proxy configuration, pairing flows, and authentication options in detail.

Two Ports, Two Separate Processes (Port 9119 vs 8642)

Hermes exposes two distinct HTTP interfaces via two separate processes:

• Port 9119 — the workspace / dashboard (hermes dashboard). A browser UI for monitoring, configuration, cron management, skills, logs, and shell-exec approvals. Intended for the agent operator. Requires the [web] install extra. Binds to localhost only with no built-in auth.
• Port 8642 — the OpenAI-compatible API server (hermes gateway). Accepts POST /v1/chat/completions, /v1/responses, /v1/models, /v1/runs, and /health calls. Intended for programmatic integrations — any client that speaks the OpenAI API protocol can point at this port.

The two processes can run side by side on the same host. They are independent: starting or stopping the dashboard does not affect the gateway, and vice versa.

The API server is not enabled by default. To activate it, set API_SERVER_ENABLED=true in your ~/.hermes/.env file and restart the gateway. (Config.yaml support for this setting is planned for a future release but is not available yet.)

If you are looking for an endpoint to integrate with an external tool, you want port 8642 via the gateway. If you are looking to monitor what your agent is doing or approve a pending action, you want the workspace at port 9119 via hermes dashboard.

Security Notes

The workspace has broad access to the agent: it can modify configuration, approve shell commands, manage skills, and inspect all sessions. Treat access to it like access to the machine itself.

• Do not expose port 9119 directly to the public internet without TLS and authentication. An unauthenticated workspace means anyone who can reach the port can approve arbitrary shell commands.
• Use a reverse proxy with auth. At minimum, put a TLS-terminating proxy in front of the workspace. Add HTTP Basic Auth or OAuth-level access control if the workspace domain is publicly reachable.
• Keep the gateway auth token private. The token in your Hermes config grants full control over the agent. Rotate it if you suspect it was exposed.
• Or use OpenClaw Launch’s managed gateway (see section below) and skip the port-exposure problem entirely.

Workspace on OpenClaw Launch

When you deploy a Hermes Agent instance on OpenClaw Launch, the workspace is pre-configured with TLS and session-based authentication. You reach it directly from your OpenClaw Launch dashboard without touching any ports, configuring any reverse proxy, or managing any certificates. Shell-exec approvals, skill management, cron tasks, and log viewing all work the same way — the only difference is that the infrastructure is handled for you.

Frequently Asked Questions

Is the Hermes dashboard the same as the Hermes workspace?

Yes. “Hermes dashboard” and “Hermes workspace” are used interchangeably in community discussion to refer to the same browser UI: the FastAPI + React management interface that ships with the Hermes Agent gateway.

What port does the Hermes workspace run on?

Port 9119 by default. The dashboard binds to 127.0.0.1 (localhost only) and requires running hermes dashboard separately from the gateway. The separate OpenAI-compatible API server runs on port 8642 via hermes gateway and must be enabled with API_SERVER_ENABLED=true in ~/.hermes/.env.

Can I disable the Hermes workspace?

The workspace is a separate process started by hermes dashboard. To disable it, simply do not run that command. The gateway (hermes gateway) operates independently and continues to work without the dashboard running. If you want the workspace accessible only locally, do not expose port 9119 outside the host machine and do not set up any proxy to it.

Does Hermes have a mobile app?

No. The Hermes workspace is a web UI only, accessible from any browser. There is no dedicated native mobile app. You can use it on mobile through a browser pointed at your workspace URL, provided the URL is reachable and secured with HTTPS.

Can I send messages to the agent from the workspace?

The Sessions page in the workspace is read-only — you can browse session history but not send new messages from it. To converse with the agent, use a connected platform channel (Telegram, Discord, WhatsApp) or send requests to the OpenAI-compatible API endpoint on port 8642.

Related Guides

• Hermes Agent Web UI — step-by-step setup and access, reverse-proxy configuration, and pairing flows
• Install Hermes Agent — self-hosted installation walkthrough
• What is Hermes Agent? — overview of what Hermes is and how it differs from other AI agent frameworks
• Hermes Agent vs alternatives — comparison with other self-hosted agent options
• Deploy Hermes Agent — full deployment guide including Docker, env vars, and config file setup