The OpenClaw Commands You'll Actually Use (With Explanations)
Why This Cheatsheet?
When you first start with OpenClaw, you have an agent but no clear picture of which commands you need when. The official docs exist — but they don't explain why you need certain commands in what order.
This isn't a complete manual. It's the cheatsheet we wished we had when we built our 6-agent setup.
---
Gateway Commands (almost always the first step)
The gateway is the heartbeat of OpenClaw. Without a running gateway: no channel connections, no heartbeats, no cron jobs.
```bash
# Check status — is the gateway even running?
openclaw gateway status
# Start the gateway (first time or after a reboot)
openclaw gateway start
# Restart the gateway — needed after config changes
openclaw gateway restart
# Stop the gateway
openclaw gateway stop
```
When do you need restart? Whenever you add new channels, change configuration, or update credentials. Start is for the initial launch; restart for everything after.
Tip: In production setups, the gateway runs as a systemd service — not started manually. In our playbook, we show how we set this up on our Hetzner server.
---
Channel Commands (how the agent communicates)
An agent without a channel is an agent that talks to nobody. Channels connect your agent to Telegram, Discord, WhatsApp, Signal, etc.
```bash
# Show all configured channels
openclaw channels list
# Add a new channel (interactive wizard)
openclaw channels add
# Remove a channel
openclaw channels remove <channel-name>
# Test a channel — sends a test message
openclaw channels test <channel-name>
```
The most common mistake: Adding a channel and forgetting to run `openclaw gateway restart` afterward. The agent won't see the new channel until after the restart.
Which channel first? We recommend starting with Telegram — easiest setup, best test for the entire stack. Discord for team setups, Signal when privacy is critical.
---
Agent Commands (who does what)
```bash
# Show all configured agents
openclaw agents list
# Create a new agent
openclaw agents create
# Show agent details
openclaw agents show <agent-id>
# Delete an agent
openclaw agents delete <agent-id>
```
Important for multi-agent setups: Each agent needs its own session configuration and its own channels. A mistake many people make: running all agents on the same Telegram bot. That causes confusion. Each agent gets its own bot token.
In our setup, Sam (team lead) has her own Telegram bot, Peter (coding) has his own, and so on.
---
Cron Commands (automated tasks)
Cron jobs are the difference between an agent that waits to be spoken to — and one that works proactively.
```bash
# Show all cron jobs
openclaw cron list
# Create a new cron job (interactive)
openclaw cron create
# Enable/disable a cron job
openclaw cron enable <job-id>
openclaw cron disable <job-id>
# Delete a cron job
openclaw cron delete <job-id>
# Manually trigger a cron job (for testing)
openclaw cron trigger <job-id>
```
Real example from our setup:
Sam runs a morning report daily at 9:00 AM: checks emails, calendar, and open ClickUp tasks. The cron job sends the result directly into the Telegram chat with the CEO.
Also: daily at 6:00 AM, Sam automatically publishes a blog post to our website — triggered by a cron job.
Timing format: OpenClaw uses standard cron syntax (`0 9 * * 1-5` = Monday–Friday 9:00 AM).
---
Config Commands
```bash
# Show full configuration
openclaw config show
# Set a single value
openclaw config set <key> <value>
# Check OpenClaw version
openclaw version
```
Tip: `openclaw config show` is great for debugging. You immediately see which channels are active, which model is configured, and which API keys are set.
---
Session Commands (what's running right now?)
```bash
# Show active sessions
openclaw sessions list
# View session logs
openclaw sessions logs <session-id>
# Kill a session
openclaw sessions kill <session-id>
```
When do you need this? When an agent is hanging, a session is consuming too much memory, or you just want to know what's currently running.
---
The Most Important Order When Setting Up
Starting from zero, the order matters:
```
1. openclaw gateway start
2. openclaw channels add (e.g. Telegram)
3. openclaw gateway restart
4. openclaw agents create
5. Test: send a message to the bot
6. openclaw cron create (optional, for automated tasks)
```
This is the simplified version. The complete setup — including Docker, Tailscale security, workspace configuration, and multi-agent coordination — is documented in the OpenClaw Setup Playbook. 18 chapters, based on real production experience.
Fully available in German too. 🇩🇪
Want to learn more?
Our playbook contains 18 detailed chapters — available in English and German.
Get the Playbook