1. Browser Didn't Open After Install
After the OpenClaw installer finishes, the dashboard should open automatically. If it doesn't, don't worry — your install is likely fine. Just open your browser manually and go to:
That's the OpenClaw dashboard. If it loads, you're good to go.
Why didn't it open automatically?
The installer runs in a restricted macOS context that can't always launch a browser window. This is a known macOS limitation — not an install failure.
What if the page won't load?
If the browser gives you a "connection refused" or blank page, the Gateway may not have started yet. Give it 10–15 seconds after install and try again.
If it still won't load, check the Gateway status:
openclaw gateway status
If it says the gateway is not running, start it:
openclaw gateway start
Then try http://127.0.0.1:18789/chat again.
I get a "token" or "unauthorized" error on the page
Your browser may have a stale token cached from a previous install. Try opening the URL in an incognito/private window instead.
Or, generate a fresh dashboard URL from the terminal:
openclaw dashboard
This opens the browser with a fresh token pre-populated.
2. Gateway Won't Start or Gateway Error
The Gateway is the local server that runs in the background and powers everything in OpenClaw. If it's not running, the dashboard won't load and your agent won't respond.
Check the Gateway status
openclaw gateway status
This shows whether it's running, what port it's on, and where the logs are.
Start or restart the Gateway
openclaw gateway start
openclaw gateway restart
View the logs
If the gateway keeps failing to start, check the logs for the error:
openclaw gateway status
The log file path is listed in the output (e.g. /tmp/openclaw/openclaw-2026-03-17.log). Open it with:
tail -50 /tmp/openclaw/openclaw-$(date +%Y-%m-%d).log
Port already in use
If you see a port conflict error, another process is using port 18789. You can either stop that process or run the gateway on a different port:
openclaw gateway start --port 18790
Run openclaw status for a full health check
openclaw status
This gives you a full picture of what's configured, what's running, and what's missing.
3. API Key or Model Error
If OpenClaw can't reach your AI provider, you'll see errors like "authentication failed", "invalid API key", or "model not found". Here's how to fix it.
Check your current model and auth status
openclaw models list
This shows all configured models, which one is the default, and whether auth is set up for each.
Set your default model
openclaw models set anthropic/claude-opus-4-5
Replace anthropic/claude-opus-4-5 with whichever model you want to use. Run openclaw models list to see available options.
Add or update your API key
openclaw models auth add
This walks you through an interactive prompt to select your provider and paste your API key. It handles Anthropic, OpenAI, Google, and others.
Where to get your API key
- Anthropic (Claude): console.anthropic.com → API Keys
- OpenAI (GPT): platform.openai.com → API Keys
- Google (Gemini): aistudio.google.com → Get API Key
Still getting errors?
Run the full health check — it will flag any missing or misconfigured auth:
openclaw status