OpenClaw Command Reference: Setup, Gateway, and Logs

After my OpenClaw deployment guide (Chinese original), I collected the console commands I use most often. Commands and options can change; openclaw --help and the relevant subcommand’s --help show what your installed version supports.

Install and configure

The original Windows PowerShell installer downloads and runs the project’s script. Review it before execution:

powershell -c "irm https://openclaw.ai/install.ps1 | iex"

Configure model credentials, channels, and other settings:

openclaw configure

Initialize a workspace or use the onboarding wizard:

openclaw setup
openclaw onboard

Updates and the gateway

openclaw update
openclaw gateway start
openclaw gateway stop
openclaw gateway restart
openclaw gateway status

Gateway start controls an installed gateway service; first-time onboarding may need to install that service. Status helps check whether it is running.

Troubleshoot or send a message

openclaw doctor
openclaw agent --message "Hello"

Doctor performs diagnostics. Review any proposed repairs before applying them. The agent command supports additional options; in Telegram, commands such as /model and /think can select a model or thinking level where supported.

Sessions, memory, schedules, and logs

openclaw sessions
openclaw memory status
openclaw cron list
openclaw cron status
openclaw logs

Use openclaw cron --help for adding and managing scheduled tasks. Logs are helpful when investigating a failed connection; remove tokens and private messages before sharing them.

The official CLI reference and gateway reference provide the current options. Read release notes as well as updating regularly.

Adapted from the original Chinese article, published on April 8, 2026.