What's already set up
After you deploy OpenClaw in one click, the Gateway container is running and bound to port 18789, with its config and memory stored in a persistent ~/.openclaw volume so upgrades don't wipe your data. No messaging channel and no AI model provider are configured yet — both are chosen during the onboarding wizard on first login, not preset by the template.
First login
Open the Control UI in your browser at http://<server_IP>:18789/ — or via the access link shown in your Leneex panel, if the template puts a domain and SSL in front of it. You'll need a gateway access token to unlock the dashboard; it's generated during setup and written to the .env file inside the OpenClaw installation directory. Paste the token into Settings, then complete onboarding: pick a model provider (a cloud API key, or point it at a local Ollama instance), and connect your first channel — for Telegram, that means creating a bot via @BotFather and pasting its token into the wizard.
Connecting channels and staying safe
Each additional channel (WhatsApp, Discord, Slack, Matrix, and others) is added the same way after initial setup, one at a time, through either the Control UI or the CLI. Because OpenClaw executes real terminal commands and can install skills from a community registry, review a skill's source before enabling it — some published skills have been found to send data out without the user noticing.
Keep the Gateway image current with regular updates, since the project releases frequently and security fixes land in new builds rather than being backported.