Appendix F: Command Quick Reference
Just installed OpenClaw and feeling overwhelmed by dozens of commands? This quick reference organizes all CLI commands by use case — from installation to daily operations, from channel management to troubleshooting. Look up what you need, when you need it — no need to read through everything.
Reading tip: Commands marked with ⭐ are the most frequently used and you'll need them right after installation. Commands marked with 🔧 are advanced operational commands that you may not encounter in everyday use. The inline comment for each command explains its purpose.
Quick Navigation
| Scenario | Jump To | One-Line Summary |
|---|---|---|
| Just installed, want to try it | Installation & Updates | Install, upgrade, uninstall |
| First-time setup | Initialization & Configuration | Wizard, diagnostics, Dashboard |
| Chat directly in the terminal | Terminal Chat | TUI interactive chat, single Agent call |
| Get OpenClaw running | Gateway Management | Start, stop, health check |
| Connect to chat platforms | Channel Management | Add Telegram / QQ / Feishu and other channels |
| Manage multiple Agents | Agent Management | Create, bind, run Agents |
| Install skills | Skills & Plugins | Browse, install, enable skills and plugins |
| Select and switch models | Model Management | Set model, fallback, aliases |
| Schedule automatic execution | Scheduled Tasks | Add, edit, delete, view Cron jobs |
| Send messages, polls | Message Sending | Proactively send messages and interactions |
| Tune parameters | Configuration Operations | Read and write configuration values |
| Sandbox and security | Sandbox & Security | Sandbox isolation, approvals, security audits |
| Something went wrong | Diagnostics & Debugging | Logs, Doctor, backup and restore |
| Other advanced features | More Commands | Memory, devices, browser, DNS, Webhooks, etc. |
Global Options
All openclaw commands support the following options:
openclaw [--dev] [--profile <name>] [--log-level <level>] <command>
--dev # Isolated development environment (state stored in ~/.openclaw-dev, ports auto-offset)
--profile <name> # Custom isolated environment (state stored in ~/.openclaw-<name>)
--log-level <level> # Override global log level (silent|fatal|error|warn|info|debug|trace)
--no-color # Disable terminal color output
-V, --version # Show version numberMultiple environments stay isolated from each other:
--devis for debugging,--profileis for running multiple instances simultaneously (e.g., one production, one test).
Installation & Updates
⭐ Install OpenClaw
Windows users (open PowerShell as Administrator):
Set-ExecutionPolicy -ExecutionPolicy RemoteSigned -Scope CurrentUser
iwr -useb https://openclaw.ai/install.ps1 | iexThe one-click script automatically installs Node.js and OpenClaw, then launches the configuration wizard immediately after.
macOS / Linux / WSL2 users:
npm install -g openclaw@latest # Install via npm (recommended)
# or
pnpm add -g openclaw@latest # Install via pnpmPrerequisite: Node.js >= 22. See Chapter 2: Manual Installation.
Update & Uninstall
openclaw update # ⭐ Update to the latest version
openclaw update --channel stable|beta|dev # Switch to a specific update channel
openclaw --version # Show the current version
openclaw uninstall # Uninstall gateway service and local data (keeps CLI)
openclaw uninstall --all # Completely uninstall
openclaw reset # Reset configuration and state (keeps CLI)
openclaw reset --scope config+creds+sessions --yes # Precisely reset specified scopeInitialization & Configuration
The first step after installation: run the configuration wizard.
⭐ Configuration Wizard
openclaw onboard # Interactive setup wizard (recommended for first use)
openclaw onboard --install-daemon # Set up and install daemon (start on boot)
openclaw onboard --non-interactive --mode local # Non-interactive mode (for scripted deployment)
openclaw onboard --reset # Re-run the wizard (overwrites existing configuration)⭐ Configuration Validation
Not sure if your configuration is correct? Use Doctor for a one-click diagnosis:
openclaw doctor # Automatically diagnose common issues
openclaw doctor --fix # Diagnose and auto-fix
openclaw doctor --deep # Deep scan (checks more items)Reconfigure
openclaw configure # Reopen the interactive configuration wizard⭐ Open Dashboard
openclaw dashboard # Launch the Web Dashboard (browser opens automatically)Terminal Chat
No browser needed, no chat platform required — talk to OpenClaw directly in the terminal. SSH remote, servers, Docker containers — if there's a terminal, you can chat.
⭐ TUI Interactive Mode
openclaw tui # Open the terminal chat interface (connects to Gateway)Single Agent Call
No interactive interface needed — just send one message and go:
openclaw agent --message "Summarize today's schedule for me" # Execute locally
openclaw agent --message "Hello" --to +15555550123 --deliver # Execute and deliver to chat channel
openclaw agent --message "Analyze report" --thinking high # Use deep thinkingSearch Documentation
openclaw docs # Search OpenClaw official documentationGateway Management
The Gateway is OpenClaw's core backend service — all message sending and receiving, skill calls go through it. Default address: ws://127.0.0.1:18789.
⭐ Daily Management
openclaw gateway status # Check if the gateway is running
openclaw gateway start # Start the gateway service
openclaw gateway stop # Stop the gateway service
openclaw gateway restart # Restart (required after changing configuration)Startup Options
Use these when you need to customize the startup method:
openclaw gateway --port 18789 --verbose # Run in foreground (debug mode, logs output directly)
openclaw gateway --bind loopback # Local access only (default, most secure)
openclaw gateway --bind lan # Local network access
openclaw gateway --bind tailnet # Tailscale network access
openclaw gateway --auth token --token <t> # Token authentication
openclaw gateway --auth password --password <p> # Password authentication
openclaw gateway --tailscale serve|funnel # Expose service via Tailscale
openclaw gateway --dev # Development modeService Installation
openclaw gateway install # Install as a system service (start on boot)
openclaw gateway uninstall # Uninstall the system serviceHealth Check
openclaw health # 🔧 Gateway health status
openclaw health --json # JSON format output (suitable for script parsing)
openclaw health --verbose # Detailed information
openclaw gateway call <method> --params '<json>' # 🔧 Directly call a gateway RPC methodChannel Management
Channels are the chat platforms OpenClaw connects to — Telegram, QQ, Feishu, Discord, WhatsApp, and more.
⭐ Add a Channel
openclaw channels add # Interactive add (recommended, with guided prompts)
# Non-interactive add examples:
openclaw channels add \
--channel telegram \
--account alerts \
--name "Alerts Bot" \
--token $TELEGRAM_BOT_TOKEN
openclaw channels add \
--channel discord \
--account work \
--name "Work Bot" \
--token $DISCORD_BOT_TOKENView Status
openclaw channels list # ⭐ List all added channels
openclaw channels list --json # JSON format
openclaw channels status # View connection status of each channel
openclaw channels status --probe # Actively probe connections (more accurate)
openclaw channels logs # View channel logs
openclaw channels logs --channel <name> --limit 200 # Specify channel, limit log countRemove a Channel
openclaw channels remove --channel <channel> --account <id>
openclaw channels remove --channel <channel> --account <id> --delete # Also delete dataLogin / Logout (platforms requiring QR code scan, e.g. WhatsApp)
openclaw channels login # Open QR code login
openclaw channels login --channel whatsapp --account <id> --verbose
openclaw channels logout # Log out
openclaw channels logout --channel <channel> --account <id>⭐ Pairing Management
Pairing is OpenClaw's security mechanism — new users require approval on their first conversation.
openclaw pairing list # View pending pairing requests
openclaw pairing approve <channel> <code> # Approve a pairing request
openclaw pairing approve --channel <channel> <code> --notify # Approve and notify the other partyAgent Management
Agents are OpenClaw's intelligent entities — each Agent has its own model, workspace, and bound channels.
View Agents
openclaw agents list # ⭐ List all Agents
openclaw agents list --json # JSON format
openclaw agents list --bindings # Show which channels each Agent is bound toAdd an Agent
openclaw agents add # ⭐ Interactive add (recommended)
openclaw agents add [name] # Specify a name
# Non-interactive (for automated deployment):
openclaw agents add my-agent \
--workspace ~/.openclaw/workspace-my-agent \
--model anthropic/claude-sonnet-4-5 \
--bind whatsapp:personal \
--non-interactiveBind Channels
One Agent can be bound to multiple channels, and one channel can be shared by multiple Agents.
openclaw agents bindings # View all binding relationships
openclaw agents bindings --agent <id> # View bindings for a specific Agent
openclaw agents bind --agent <id> --bind <channel[:accountId]> # Add a binding
openclaw agents unbind --agent <id> --bind <channel[:accountId]> # Remove a binding
openclaw agents unbind --agent <id> --all # Remove all bindingsRun an Agent (manual trigger)
openclaw agent --message "Hello" --to <dest> # Send a message
openclaw agent --message "Hello" --session-id <id> # Send in a specific session
openclaw agent --message "Hello" --thinking high # Use deep thinking
openclaw agent --message "Hello" --local # Local mode (bypasses gateway)
openclaw agent --message "Hello" --channel whatsapp --deliver # Send and deliver to channelDelete an Agent
openclaw agents delete <id> # Delete an Agent
openclaw agents delete <id> --force # Force delete (skip confirmation)Skills & Plugins
Skills extend OpenClaw's capabilities — search, scheduling, smart home, and more are all powered by skills. Plugins are a lower-level extension mechanism.
Skill Management
openclaw skills list # ⭐ List installed skills
openclaw skills list --eligible # Show only available skills (dependencies satisfied)
openclaw skills list -v # Show details (including missing dependencies)
openclaw skills list --json # JSON format
openclaw skills info <name> # View skill details
openclaw skills check # Check skill health statusTo install new skills, use the
clawhubCLI, notopenclaw skills. See Appendix D: Skill Development & Publishing Guide.
# Common clawhub commands
npm i -g clawhub # Install clawhub CLI
clawhub search <keyword> # Search for skills
clawhub install <skillname> # Install a skill
clawhub list # List installed skills
clawhub update --all # Update all skills
clawhub uninstall <skillname> # Uninstall a skillPlugin Management
openclaw plugins list # List installed plugins
openclaw plugins list --json # JSON format
openclaw plugins info <id> # View plugin details
openclaw plugins install <path|.tgz|npm-spec> # Install a plugin
openclaw plugins enable <id> # Enable a plugin
openclaw plugins disable <id> # Disable a plugin
openclaw plugins doctor # Plugin diagnosticsModel Management
OpenClaw supports multiple model providers, using the provider/model-name format to identify models.
⭐ Common Operations
openclaw models list # List available models
openclaw models status # View the currently used model and its status
openclaw models set <model> # Switch the default text model
openclaw models set-image <model> # Switch the default image modelModel Aliases
Find model names too long? Set up aliases:
openclaw models aliases list # View alias list
openclaw models aliases add <alias> <model> # Add an alias (e.g., gpt → openai/gpt-4o)
openclaw models aliases remove <alias> # Remove an aliasModel Fallback
Automatically switch to a backup model when the primary model is unavailable:
openclaw models fallbacks list # View the fallback chain
openclaw models fallbacks add <model> # Add a fallback model
openclaw models fallbacks remove <model> # Remove a fallback model
openclaw models fallbacks clear # Clear the fallback chainModel Authentication
openclaw models auth add # 🔧 Add model provider authentication
openclaw models auth setup-token # 🔧 Set up a Token
openclaw models auth paste-token # 🔧 Paste a Token
openclaw models auth order get|set|clear # 🔧 Manage provider priorityFor model configuration details, see Chapter 5: Model Management. For provider comparisons, see Appendix E: Model Provider Selection Guide.
Scheduled Tasks
Have OpenClaw automatically execute tasks on a time schedule — morning briefings, timed reminders, data checks, and more.
⭐ Common Operations
openclaw cron list # View all scheduled tasks
openclaw cron status # Overview of task execution status
# Add a task (three trigger methods):
openclaw cron add --name "Morning Briefing" --cron "0 8 * * *" --message "Today's weather and schedule" --channel "telegram:chat:123"
openclaw cron add --name "Timed Reminder" --every 10m --message "Time to drink water"
openclaw cron add --name "Delayed Task" --at 20m --message "Remind me in 20 minutes"Manage Tasks
openclaw cron edit <id> # Edit a task
openclaw cron rm <id> # Delete a task (note: rm, not delete)
openclaw cron enable <id> # Enable a task
openclaw cron disable <id> # Disable a task
openclaw cron run <id> # Manually trigger a task immediately
openclaw cron runs # View execution historyChannel Format Reference
The --channel parameter format varies by platform:
telegram:chat:<ChatID>
qqbot:c2c:<openid>
qqbot:group:<groupid>
feishu:chat:<ChatID>
discord:channel:<ChannelID>Message Sending
Send messages directly from the terminal, without going through a chat platform:
Send a Message
openclaw message send --target +15555550123 --message "Hello"Send a Poll
openclaw message poll \
--channel discord \
--target channel:123 \
--poll-question "What should we have for lunch?" \
--poll-option "Hot pot" \
--poll-option "Sushi"Other Message Operations
openclaw message react # Add an emoji reaction
openclaw message reactions # View reactions
openclaw message read # Mark as read
openclaw message edit # Edit a message
openclaw message delete # Delete a message
openclaw message pin # Pin a message
openclaw message unpin # Unpin a messageConfiguration Operations
OpenClaw's configuration can be read and written directly via CLI (for configuration file structure, see Appendix G).
openclaw config get <path> # Read a configuration value
openclaw config get agents.defaults.workspace # Example: read the default workspace path
openclaw config set <path> <value> # Set a configuration value
openclaw config set agents.defaults.heartbeat.every "2h" # Example: set heartbeat interval
openclaw config set tools.profile full # Example: set tool configuration
openclaw config set logging.level debug # Example: enable debug logging
openclaw config unset <path> # Delete a configuration value
openclaw config file # View the configuration file path
openclaw config validate # Validate the configuration
openclaw config validate --json # Output validation results in JSON formatFor the complete configuration file reference, see Appendix G.
Diagnostics & Debugging
When something goes wrong, troubleshoot in this order: Doctor → Logs → Status → Security Audit.
⭐ Step 1: Doctor Diagnosis
openclaw doctor # Automatically detect common issues
openclaw doctor --fix # Detect and auto-fix
openclaw doctor --deep # Deep scanView Logs
openclaw logs --follow # ⭐ Stream logs in real time (like tail -f)
openclaw logs --limit 100 # Last 100 log entries
openclaw logs --limit 100 --json # JSON format (easier to search)
openclaw logs --limit 50 --plain # Plain text formatView System Status
openclaw status # ⭐ Overview of running status
openclaw status --deep # Deep health check
openclaw status --json # JSON format
openclaw status --usage # Include usage information🔧 Security Audit
openclaw security audit # Security audit
openclaw security audit --deep # Deep security scan
openclaw security audit --fix # Detect and auto-fix security issues🔧 Secret Management
openclaw secrets reload # Reload secrets
openclaw secrets audit # Secret security audit
openclaw secrets configure # Configure secrets
openclaw secrets apply --from <plan.json> # Batch apply from a plan fileFor security-related details, see Chapter 10: Security Protection & Threat Model.
Sandbox & Security
Sandbox Management
Sandboxes provide an isolated runtime environment for Agents, preventing operations from affecting the host system:
openclaw sandbox list # List sandbox containers
openclaw sandbox start # Start a sandbox
openclaw sandbox stop # Stop a sandboxFor sandbox details, see Chapter 8: Gateway Operations and Chapter 10: Security Protection.
Execution Approvals
Agent execution of sensitive operations requires human approval:
openclaw approvals list # View pending approval operations
openclaw approvals approve <id> # Approve
openclaw approvals deny <id> # Deny🔧 Agent Hooks
openclaw hooks list # List registered Agent Hooks
openclaw hooks add # Add a Hook
openclaw hooks remove <id> # Remove a HookMore Commands
The following commands are used less frequently — consult as needed.
Memory Management
OpenClaw automatically remembers conversation context. These commands are for manual memory management:
openclaw memory status # Memory system status
openclaw memory index # Manually trigger memory indexing
openclaw memory search "<query>" # Search memory contentsDevice Management
Remote devices (phones, tablets, etc.) require approval when connecting to OpenClaw:
openclaw devices list # List all devices
openclaw devices list --json # JSON format
openclaw devices approve [requestId] # Approve a device
openclaw devices approve --latest # Approve the latest request
openclaw devices reject <requestId> # Reject a device
openclaw devices remove <deviceId> # Remove an authorized device
openclaw devices clear --yes # Clear all devices
openclaw devices clear --yes --pending # Clear only pending devicesBrowser Management
OpenClaw has a built-in headless browser for web operation skills:
openclaw browser status # Browser status
openclaw browser start # Start the browser
openclaw browser stop # Stop the browser
openclaw browser reset-profile # Reset browser profile
openclaw browser tabs # View open tabs
openclaw browser open <url> # Open a web page
openclaw browser screenshot # Take a screenshot
openclaw browser navigate <url> # Navigate to a specified URLSession Management
openclaw sessions list # List all sessions
openclaw sessions --json # JSON format
openclaw sessions --verbose # Detailed informationSystem Management
openclaw system event # View system events
openclaw system heartbeat last # Time of the last heartbeat
openclaw system heartbeat enable # Enable heartbeat
openclaw system heartbeat disable # Disable heartbeat
openclaw system presence # Online presence statusBackup & Restore
openclaw backup create # Create a local backup archive
openclaw backup verify # Verify backup integrity
openclaw backup list # List existing backupsNode Management (Headless Nodes)
Run headless nodes on remote servers, managed and scheduled by the gateway:
openclaw node start # 🔧 Start the Node Host service
openclaw node stop # 🔧 Stop the Node Host
openclaw nodes list # 🔧 List all nodes managed by the gateway
openclaw nodes pair # 🔧 Pair a new nodeDNS & Network Discovery
openclaw dns status # 🔧 DNS discovery status (Tailscale + CoreDNS)Webhooks
openclaw webhooks list # 🔧 List configured Webhooks
openclaw webhooks add # 🔧 Add a Webhook
openclaw webhooks remove <id> # 🔧 Remove a WebhookContacts & Group Lookup
openclaw directory self # Look up your own ID
openclaw directory peers # Look up contact IDs
openclaw directory groups # Look up group IDsACP (Agent Control Protocol)
openclaw acp status # 🔧 ACP protocol statusShell Auto-completion
openclaw completion # Generate shell completion script (supports bash/zsh/fish)iOS Pairing
openclaw qr # Generate iOS pairing QR codeUsage Statistics & Cleanup
openclaw usage today # Today's token usage
openclaw usage month # Monthly usage
openclaw usage --by-skill # Usage breakdown by skill
openclaw cleanup --conversations --older-than 7d # Clean up old conversations
openclaw cleanup --skill-cache # Clear skill cache
openclaw cleanup --older-than 30d # Clean up data older than 30 daysShortcut Aliases
Set up short aliases for frequently used commands to avoid typing long commands repeatedly:
# Add to ~/.bashrc or ~/.zshrc
alias oc='openclaw'
alias oct='openclaw tui'
alias ocg='openclaw gateway'
alias ocs='openclaw skills'
alias oca='openclaw agents'
alias ocd='openclaw dashboard'
alias ocdr='openclaw doctor'
alias occh='openclaw channels'
# One-command start/stop
alias ocstart='openclaw gateway start && openclaw dashboard'
alias ocstop='openclaw gateway stop'
alias ocrestart='openclaw gateway restart'Troubleshooting Quick Reference
| Symptom | Try This First | If That Doesn't Work |
|---|---|---|
| Gateway fails to start | openclaw doctor --fix | Check port conflicts: lsof -i :18789 |
| Dashboard won't open | openclaw gateway status | Confirm gateway is running, check firewall |
| Channel won't connect | openclaw channels status --probe | Check if Token has expired, view channel logs |
| Model call fails | openclaw models status | Verify API Key, check balance |
| Device pairing fails | openclaw devices list | Approve pending requests |
| Configuration not taking effect | openclaw config validate | Restart gateway: openclaw gateway restart |
| TUI can't connect | openclaw gateway status | Confirm gateway is running, check port |
| Nothing works | openclaw doctor --deep | View logs: openclaw logs --limit 50 |
Getting Help
Can't remember a command? Add --help anywhere:
openclaw --help # All top-level commands
openclaw <command> --help # Subcommands for a specific command
openclaw channels add --help # Parameter details for a specific commandNote: This quick reference is compiled from the OpenClaw official CLI Reference. Commands may change with version updates — visit the OpenClaw official documentation for the latest information.