Skip to main content

​
Getting started

This page takes you from zero to a running WednesdayAI gateway with the control panel accessible. Channel setup — connecting WhatsApp, Telegram, iMessage, and others — comes after the gateway is stable.
Fastest path to a first chat: once the gateway is running, open http://127.0.0.1:18789/ in a browser on the same machine. The built-in control panel lets you chat with your AI without connecting any external channel first.

​
Requirements

  • Node.js 24 or newer — the installer script installs it automatically if missing
  • macOS, Linux, or Windows (via WSL2) — native Windows is supported but WSL2 is recommended
  • About 500 MB free disk space (including node modules)
Check your Node version: 
node --version
If it prints v24.x.x or higher you are ready. Otherwise let the installer handle it, or install via nvm / fnm first.
WhatsApp and Telegram require the gateway to run on Node, not Bun. Version-managed Node paths (nvm, fnm, volta) can break a background service after upgrades — openclaw doctor warns and offers to migrate to a system Node install.

​
Step 1 — Install

The installer script is the recommended path. It detects your Node version, installs or upgrades it if needed, installs the openclaw CLI globally, and launches the onboarding wizard. 
curl -fsSL https://wednesdayai.dev/install.sh | bash
pnpm needs an extra step to approve packages with native build scripts:
pnpm add -g openclaw@latest
pnpm approve-builds -g    # select openclaw, sharp, node-llama-cpp when prompted
If sharp fails because you have libvips installed via Homebrew:
SHARP_IGNORE_GLOBAL_LIBVIPS=1 npm install -g openclaw@latest

​
Step 2 — Run the onboarding wizard

If the installer completed successfully it already ran the wizard. After a manual npm install, run: 
openclaw onboard --install-daemon
The wizard walks you through:
  1. Model and auth — choose an AI provider and authenticate. It supports Anthropic (API key or claude setup-token) and OpenAI Codex (OAuth) among others. You can add more later via openclaw onboard or openclaw models auth ....
  2. Gateway port — defaults to 18789. Change it here if that port is in use.
  3. System service — --install-daemon installs the gateway as a background service so it starts at login. Recommended for persistent deployments.
Rerun the wizard at any time with openclaw onboard. Provider auth can also be added directly: openclaw models auth paste-token --provider anthropic or openclaw models auth login --provider <provider>.

​
Step 3 — Verify the gateway

openclaw gateway status
Expected: 
✔ Gateway running  pid=12345  port=18789  uptime=0m 12s
If it is not running, start it: 
systemctl --user start openclaw-gateway
systemctl --user status openclaw-gateway

​
Step 4 — Open the control panel

openclaw dashboard
This opens http://127.0.0.1:18789/. If the gateway runs on a remote server, see Remote access.
If the control panel loads and you can send a test message, your gateway is ready.

​
Step 5 — Run a health check

openclaw doctor
openclaw doctor validates your config, checks configured channels and providers, and reports missing credentials or unknown config keys. Fix any warnings before moving on. 
openclaw doctor --fix    # auto-repair common issues
See Doctor for the full check list.

​
What you have now

ComponentStatus
openclaw CLIInstalled globally
GatewayRunning as a background service
AuthAt least one AI provider connected
Control panelAccessible at http://localhost:18789/

​
Next steps

Gateway configuration

Tune port, bind, auth, model routing, and all other gateway settings.

Connect channels

Add WhatsApp, Telegram, iMessage, Signal, Slack, Discord, and more.

AI providers

API keys, OAuth, and model selection.

Remote access

Reach the gateway securely from outside localhost.

​
Troubleshooting

The npm global bin directory is not in your PATH:
echo 'export PATH="$(npm prefix -g)/bin:$PATH"' >> ~/.zshrc
source ~/.zshrc
openclaw --version
Change the port in ~/.openclaw/openclaw.json:
{ "gateway": { "port": 18790 } }
Then restart: systemctl --user restart openclaw-gateway (Linux) or openclaw gateway start (macOS/manual).
The wizard requires an active TTY. Over SSH without a PTY, add -t:
ssh -t user@host openclaw onboard
Or complete the wizard locally and copy ~/.openclaw/openclaw.json to the server.
Unknown top-level keys stop the gateway from starting. Remove them, or run openclaw doctor --fix (check the diff before restarting).