Dashboard Guide
OpenClaw Dashboard: Managing Your Bots, Models and Channels
“OpenClaw dashboard” can mean two things: the Control UI that ships with OpenClaw and opens with openclaw dashboard, or the OpenClaw Launch dashboard where hosted bots are deployed and managed. This guide covers both, and walks through what each part of the Launch dashboard does.
Two dashboards, one name
| OpenClaw Control UI | OpenClaw Launch dashboard | |
|---|---|---|
| What it is | The browser admin page built into the OpenClaw Gateway | The web app at openclawlaunch.com/dashboard |
| Scope | One Gateway: chat, config, exec approvals | All your bots: deploy, models, platforms, files, billing |
| How you open it | openclaw dashboard, or http://127.0.0.1:18789/ | Sign in with Google, GitHub or email |
| Who runs the server | You | We do |
They are not competitors. Every OpenClaw bot on OpenClaw Launch still has its own Control UI, reachable from the Launch dashboard with one click.
The built-in OpenClaw dashboard (self-hosted)
If you installed OpenClaw yourself, the dashboard is the Gateway’s Control UI. After onboarding the CLI opens it automatically. To reopen it later:
openclaw dashboard # open the Control UI in your browser
openclaw dashboard --no-open # print the link insteadIt needs a running Gateway, and by default it is served at http://127.0.0.1:18789/. Authentication happens when the page connects to the Gateway, using the Gateway token or password (or Tailscale identity if you enabled that). The upstream docs are clear that this is an admin surface: keep it on localhost, behind Tailscale Serve, or behind an SSH tunnel, not on a public port. If it asks for a token you don’t have, see gateway token missing; for the Gateway itself, see OpenClaw Gateway.
What the Control UI does not do is run the machine. Installing, updating, keeping the process alive, backups and connecting platforms from scratch are still jobs for you and the CLI.
The OpenClaw Launch dashboard
Signing in
Sign in at openclawlaunch.com with Google, GitHub or email and password. The dashboard opens on your Instance page, which lists every bot on your account, newest first. If you have none yet, the same page is where you deploy one.
The instance list
Each bot is a card with a status label: starting, running, sleeping, stopped or an error. The buttons change with the status:
- Start, Stop and Restart for the bot’s container.
- Wake Up for a bot that went to sleep.
- Update when a newer OpenClaw version is available for that bot.
- Revert to last working version if an OpenClaw bot crashes after a change.
- Delete to remove the bot.
You can rename a bot, and each card shows how much storage it is using against your plan’s limit.
Chatting and opening the bot
Quick Chat opens a chat panel with the bot right inside the dashboard, useful for testing before any platform is connected (more in Quick Chat). Open Gateway on an OpenClaw card opens that bot’s Control UI, the same one described above, already authenticated.
Picking models
The card has rows for the bot’s Text model, its Image model and its Search engine, each a dropdown. Changing one writes the bot’s config for you, so there is no JSON to edit. Managed models run on your plan’s credits; if you add your own provider keys on the API Keys page, those providers show up in the picker too. On Hermes bots a model switch restarts the bot for a few seconds, and the card says so while it happens. For choosing a model, see best OpenClaw models.
Connecting channels
Connect Platforms lists the chat apps a bot can join:
| Platform | OpenClaw | Hermes |
|---|---|---|
| Telegram | Yes | Yes |
| Discord | Yes | Yes |
| Yes | Yes | |
| Yes | Yes | |
| Feishu | Yes | Yes |
| Slack | No | Yes |
| OpenClaw App | Yes | No |
| Hermes Desktop | No | Yes |
Telegram uses pairing, so nobody can talk to your bot until you approve them: the bot gives you a pairing code, and you paste it into the dashboard to approve. The card then shows the platform as paired. The walkthrough is in Telegram pairing.
Files, Terminal and Live Browser
- Files opens a file browser on the bot’s workspace, so you can read and edit what it has written. It is on the Lite and Pro plans.
- Terminal opens a shell inside the bot’s container. This is where you go to check something the card does not show.
- Live Browser shows the browser the bot drives, so you can watch it work or sign in to a site for it.
The sidebar
Beyond the instance list, the sidebar holds account-wide pages: API Keys for your own provider keys, Skills to add skills to a bot, Integrations for connected apps, Apps, Credits, Billing and Settings. Depending on your account you may also see Tools, which installs open-source runtimes and MCP servers into a bot (see the Tools guide), Website for sites your bot hosts, and Build & Resell for running bots for your own customers.
OpenClaw and Hermes in one dashboard
OpenClaw Launch runs two agent frameworks: OpenClaw and Hermes Agent. Both appear in the same list and share the same model rows, Connect Platforms, Quick Chat, Files and Terminal. The differences are small: a Hermes card has Open Dashboard in place of Open Gateway, which opens Hermes’s own web dashboard (covered in Hermes Agent Web UI), and the platform list differs as in the table above.
Dashboard versus self-hosting without one
Self-hosted OpenClaw gives you the Control UI and the CLI, which is enough to run one bot if you are comfortable on a server. Everything around the Gateway is manual: provisioning the machine, installing and upgrading, restarting after a crash, wiring platform tokens into config, and keeping the Control UI off the public internet. The Launch dashboard takes over that outer layer and leaves the agent itself unchanged. For a fuller trade-off, see managed vs self-hosted, or compare providers in best OpenClaw hosting and best Hermes hosting.
OpenClaw Dashboard FAQ
How do I open the OpenClaw dashboard?
On a machine where you run OpenClaw yourself, run openclaw dashboard. It opens the Gateway’s Control UI in your browser; add --no-open to print the link instead. Locally the Control UI lives at http://127.0.0.1:18789/. On OpenClaw Launch, sign in and go to /dashboard; the Open Gateway button on your bot opens the same Control UI for that bot.
Is the OpenClaw Launch dashboard the same as the OpenClaw Control UI?
No. The Control UI is part of OpenClaw itself and manages one Gateway: chat, config and exec approvals. The OpenClaw Launch dashboard sits a level above it. It creates and runs the bots, switches models, connects chat platforms, and handles plan, credits and billing. Each OpenClaw bot you run there still has its own Control UI one click away.
Can I manage Hermes Agent bots from the same dashboard?
Yes. OpenClaw and Hermes instances appear side by side in the same list. Hermes cards have Open Dashboard in place of Open Gateway, which opens Hermes’s own web dashboard, and Slack plus the Hermes Desktop app are available as connections only on Hermes bots.
Do I need the command line to use OpenClaw Launch?
No. Deploying, switching models, connecting platforms, pairing and restarting are all buttons on the dashboard. The CLI is only needed for advanced one-off work; see the OpenClaw CLI guide if you want it.
Is it safe to expose the OpenClaw Control UI to the internet?
Not directly. Upstream OpenClaw calls the Control UI an admin surface and recommends localhost, Tailscale Serve or an SSH tunnel rather than a public port. On OpenClaw Launch each bot’s gateway sits behind authentication, and you reach it from your signed-in dashboard.