Home

Switch language

Start Install GuideQuick StartDocs Overview
Docs ChannelsModels & APIGateway OpsTools & Skills
More ArticlesResourcesHelp Center
Get Started
Documentation navigation Help Center
Help Center 1 sections

FAQ

FAQ

These are the most common, practical questions right now, mostly around installation, runtime choice, Dashboard access, log locations, deployment targets and provider authentication.

Common Questions

What is the recommended installation and setup path?

For most people, install the CLI with the official script and then run openclaw onboard --install-daemon. That configures auth, Gateway, channels, services and baseline Skills in one pass.

How do I open Dashboard after onboarding?

The fastest path is openclaw dashboard. If the Gateway is already running locally, you can also open http://127.0.0.1:18789/ directly.

What runtime does OpenClaw need?

Node 22 or newer is recommended. Bun is not recommended for Gateway workloads, especially when you need WhatsApp or Telegram.

Can I run it on a VPS or Raspberry Pi?

VPS deployment is common and is often the best place for an always-on Gateway. Raspberry Pi depends on your load and model strategy, but the docs still recommend a more reliable Node host for stable operations.

Do I need to buy a Mac mini to install OpenClaw?

No. macOS is only one common path. Linux hosts and WSL2 are valid deployment routes too. macOS mainly becomes more attractive when you specifically need Apple ecosystem capabilities.

Can I switch between npm install and git/source install?

Yes. After switching, run openclaw doctor so the Gateway service entry points update to the installation method you are now using.

Where are the logs?

Start with openclaw logs --follow. File logs are commonly found in /tmp/openclaw/, and macOS LaunchAgent or Linux systemd also keep service-side logs.

Should I run the Gateway on my laptop or on a VPS?

A laptop is enough for local development and occasional use. A VPS is better when you want 24/7 uptime, multiple channels or remote access. The real question is stability and reachability, not whether the cloud is more advanced.

Do I need a Claude or OpenAI subscription?

No subscription is strictly required, but you do need at least one valid model-auth path such as OpenAI / Codex OAuth, Anthropic API keys, setup-token or another provider's API key.

Where are the config file and workspace?

Usually in ~/.openclaw/openclaw.json and ~/.openclaw/workspace. If you use --dev or --profile, those paths move into the corresponding isolated state directory.