Published: March 28, 2026 | Updated for: OpenClaw v2026.3.23 | Reading time: 16 min

Topics: OpenClaw · CLI Reference · AI Agent · Gateway · Automation · Developer Cheatsheet

OpenClaw is the fastest-growing open-source AI agent of 2026 — and its CLI is your primary control panel for everything: launching the gateway, managing channels, installing skills, scheduling cron jobs, controlling browsers, and running multi-agent workflows. This reference covers 100+ OpenClaw CLI commands with copy-paste examples, organized by workflow so you can find exactly what you need.

What Is the OpenClaw CLI?

OpenClaw (formerly Clawdbot and Moltbot) is a self-hosted, open-source personal AI agent that runs directly on your machine. It uses a WebSocket Gateway to connect large language models — including Claude, GPT, and Gemini — to messaging platforms like WhatsApp, Telegram, Discord, and Slack, and can execute shell commands, control browsers, manage files, and coordinate multi-agent workflows.

The CLI is the primary interface for all of this. Unlike most AI tools that hide complexity behind a web dashboard, OpenClaw was designed CLI-first — making it naturally compatible with DevOps pipelines, shell scripts, cron jobs, and CI automation. Every setting, every channel, every agent, every skill is manageable from the terminal.

Since going open source in early 2026, OpenClaw amassed over 325,000 GitHub stars in its first 90 days — one of the fastest-growing open-source projects in AI history. The CLI has grown to well over 100 subcommands across 40+ top-level command groups.

New to OpenClaw setup? Start with our guide on setting up OpenClaw with Claude Code before diving into this command reference.

Global Flags and Options

These flags can be appended to any openclaw command:

openclaw [global-flags] <command> [subcommand] [options]

OpenClaw Global CLI Flags
Flag	Description
`--dev`	Use the dev profile — isolates all state under `~/.openclaw-dev`. Safe for testing without affecting production.
`--profile <name>`	Use a named profile. Useful for managing separate work, personal, or client OpenClaw instances.
`--log-level <level>`	Override global log level: `silent`, `fatal`, `error`, `warn`, `info`, `debug`, `trace`
`--no-color`	Disable ANSI color output. Useful when piping to files or CI logs.
`--json`	Output results as JSON (supported on most commands). Ideal for scripting.
`-V, --version`	Print the current OpenClaw version number.
`-h, --help`	Display help and available subcommands for the current command.

Examples:

# Run in dev mode (safe sandbox)
openclaw --dev gateway start

# Use a named profile
openclaw --profile work channels list

# Get JSON output for scripting
openclaw channels list --json

# Debug mode for verbose output
openclaw --log-level debug gateway status

Setup & Onboarding Commands

Every OpenClaw journey starts here. These commands initialize your environment, create your config file, and prepare the gateway for first use.

First-Time Setup

# Interactive setup wizard (recommended for first-time users)
openclaw setup --wizard

# Full onboarding: gateway, workspace, and skills
openclaw onboard

# Setup + automatically install the background daemon
openclaw onboard --install-daemon

# Non-interactive (for Docker or CI environments)
openclaw onboard --non-interactive

# Re-run specific setup steps interactively
openclaw configure

Reset Commands

# Reset config only (preserves credentials and sessions)
openclaw reset --scope config

# Reset config + credentials + sessions (keeps workspace files)
openclaw reset --scope config+creds+sessions

# Full nuclear reset — removes everything including workspace
openclaw reset --scope full

When to use each reset scope: Use config when you've made a bad JSON edit. Use config+creds+sessions when switching AI providers. Use full only as a last resort — it wipes the workspace.

Uninstall

# Uninstall the gateway service and local data (CLI binary remains)
openclaw uninstall

Gateway Commands

The Gateway is the core WebSocket server powering OpenClaw. It handles all agent logic, message routing, tool execution, and channel communication. These are the commands you'll use most frequently.

Start, Stop, Restart

# Run the gateway in the foreground (debugging/testing)
openclaw gateway run

# Start the gateway as a background daemon service
openclaw gateway start

# Stop the daemon
openclaw gateway stop

# Restart the daemon (needed after config/plugin changes)
openclaw gateway restart

# Check if gateway is running and healthy
openclaw gateway status

# Deep status check with system-level scans
openclaw gateway status --deep

# Require the RPC connection to pass (useful in CI health checks)
openclaw gateway status --require-rpc --json

Daemon Service Management

# Install gateway as a system service (launchd/systemd/schtasks)
openclaw daemon install

# Uninstall the daemon service
openclaw daemon uninstall

# View gateway service logs
openclaw daemon logs

# Restart the daemon
openclaw daemon restart

Gateway Sessions

# List stored conversation sessions
openclaw sessions list

# List sessions as JSON for scripting
openclaw sessions list --json

Health Probe

# Fetch health from the running Gateway
openclaw health

# Verbose JSON health report
openclaw health --json --verbose

# Quick probe (no deep scan)
openclaw health --no-probe

If you're running OpenClaw on a remote server instead of locally, see our guide on deploying OpenClaw on a VPS — including security hardening and production cost optimization.

Config Commands

OpenClaw stores its entire configuration in a single JSON5 file at ~/.openclaw/openclaw.json. The config commands let you get, set, and unset values without manually editing JSON — and support hot-reload where applicable.

Read Config Values

# Get all config values
openclaw config get

# Get a specific config key
openclaw config get agents.defaults.model.primary

# Get config as JSON for scripting
openclaw config get --json

Write Config Values

# Set a specific config key
openclaw config set agents.defaults.model.primary "anthropic/claude-sonnet-4-5"

# Set maximum concurrent agents
openclaw config set agents.defaults.maxConcurrent 4

# Unset (remove) a config key
openclaw config unset agents.defaults.contextPruning

Config File Operations

# Validate the config file for syntax errors
openclaw config file validate

# View the resolved config file path
openclaw config file

# Auto-complete shell integration setup
openclaw completion

Backup & Restore

# Create a config backup
openclaw backup create

# Verify the backup is intact
openclaw backup verify

Channels Commands

Channels connect OpenClaw to the messaging platforms where you interact with your agent — Telegram, WhatsApp, Discord, Slack, Google Chat, LINE, and more. These are the commands for managing all channel connections.

View Channels

# List all connected channels
openclaw channels list

# Probe all channels for live health status
openclaw channels status --probe

# View logs for a specific channel
openclaw channels logs --channel telegram

# View logs in follow mode (live stream)
openclaw channels logs --channel discord --follow

Add Channels

# Add a Telegram bot channel
openclaw channels add --channel telegram --token YOUR_BOT_TOKEN

# Add WhatsApp (Meta Business setup — QR code scan required)
openclaw channels add whatsapp

# Add Discord bot
openclaw channels add discord --bot-token YOUR_DISCORD_TOKEN

# Add Slack
openclaw channels add --channel slack --token YOUR_SLACK_TOKEN

# Interactive login for any channel (e.g. WhatsApp QR scan)
openclaw channels login

Remove Channels

# Remove a channel connection
openclaw channels remove --channel telegram

# Remove and also delete stored credentials
openclaw channels remove --channel discord --delete

For a full WhatsApp automation walkthrough using OpenClaw channels, see our OpenClaw WhatsApp automation guide.

Models Commands

OpenClaw supports multiple AI providers — Anthropic, OpenAI Codex, GitHub Copilot, Gemini, Antigravity, and more via OpenRouter. These commands manage model discovery, configuration, and switching.

List & Status

# List all available/configured models
openclaw models list

# Check status and quota for all configured providers
openclaw models status

# Check usage/quota for a specific provider
openclaw status --usage

Set Active Model

# Set the primary model
openclaw models set "anthropic/claude-sonnet-4-5"

# Set fallback models
openclaw models fallbacks add "openai/gpt-5-mini"
openclaw models fallbacks add "openrouter/google/gemini-3-flash-preview"

# List current fallbacks
openclaw models fallbacks list

# Remove a fallback
openclaw models fallbacks remove "openai/gpt-5-mini"

Model Aliases

# List all model aliases (shortcuts for the :model slash command)
openclaw models aliases list

# Add an alias
openclaw models aliases add --model "anthropic/claude-haiku-4-5" --alias "haiku"

# Remove an alias
openclaw models aliases remove "haiku"

Agent Model Image

# Set the model image used for agent identity
openclaw models set-image --model "anthropic/claude-sonnet-4-5" --image ./avatar.png

Agents Commands

OpenClaw supports multiple isolated agents — each with its own workspace, credentials, model configuration, and routing rules. This is ideal for running separate agents for work, personal use, or different clients.

List & Inspect

# List all configured agents
openclaw agents list

# List agents with their channel bindings
openclaw agents list --bindings

# Output as JSON for scripting
openclaw agents list --json

Add & Delete Agents

# Add a new agent (interactive wizard)
openclaw agents add

# Add a new agent non-interactively
openclaw agents add --workspace ~/agents/work-agent --non-interactive

# Delete an agent
openclaw agents delete my-agent-name

Bindings — Route Channels to Agents

# List routing bindings (which channel routes to which agent)
openclaw agents bindings

# Bind a channel to an agent
openclaw agents bind --agent my-agent --channel telegram

# Unbind a channel from an agent
openclaw agents unbind --agent my-agent --channel telegram

Agent Identity

# Set agent identity from an existing identity config
openclaw agents set-identity --from-identity ./identity.json

ACP — Agent Client Protocol

# Check ACP (multi-agent coordination protocol) status
openclaw acp status

# View ACP health
openclaw acp health

Skills Commands

Skills are modular capability extensions for your OpenClaw agent — similar to plugins, but defined in YAML and designed to extend what your agent can do in conversations. They're installed from ClawHub or authored locally.

List & Inspect

# List all installed skills
openclaw skills list

# Get detailed info on a specific skill
openclaw skills info my-skill-name

# Check a skill for dependency issues
openclaw skills check my-skill-name

# Audit all skills for missing dependencies
openclaw skills audit

Writing a Custom Skill

Skills are defined as YAML files with a --- front matter block. A minimal skill structure:

---
name: my-skill
description: "What this skill does for the agent"
metadata:
  openclaw:
    requires:
      node: ">=18"
---

# Skill instructions here in Markdown
# The agent reads this file to understand how to use this skill

Place this file in your workspace and it will be auto-discovered when the gateway restarts.

Plugins Commands

Plugins are deeper integrations than skills — they hook into the gateway process itself, adding new tools, providers, or system behaviors. Most plugin changes require a gateway restart to take effect.

# List all installed plugins
openclaw plugins list

# Inspect a specific plugin
openclaw plugins inspect my-plugin-id

# Install a plugin
openclaw plugins install my-plugin-id

# Uninstall a plugin
openclaw plugins uninstall my-plugin-id

# Update all plugins
openclaw plugins update

# Enable a disabled plugin
openclaw plugins enable my-plugin-id

# Disable a plugin (without uninstalling)
openclaw plugins disable my-plugin-id

# Report plugin load errors
openclaw plugins doctor

# Browse the marketplace
openclaw plugins marketplace list

# List marketplace entries before installing
openclaw plugins marketplace list ClawHub

Important: After installing, enabling, or disabling any plugin, always run openclaw gateway restart to apply changes.

Memory Commands

OpenClaw agents have a persistent memory system. The agent can store facts, preferences, and context across sessions. Memory is backed by Markdown files in the workspace (with future support for vector-backed knowledge graphs via LanceDB/Cognee).

# Check current memory status
openclaw memory status

# Re-index all memory files
openclaw memory index

# Search memory for a specific topic
openclaw memory search "project deadlines"

Note on sensitive data: The memory system automatically excludes passwords, financial details, and health information from persistent storage. You can view and delete what the agent remembers directly from the workspace MEMORY.md file or via the dashboard.

Browser Automation Commands

OpenClaw can control Chromium-based browsers — Chrome, Brave, Edge — directly from the CLI. This enables powerful web scraping, form filling, UI testing, and web research workflows without writing any Selenium or Playwright code.

Browser Lifecycle

# Check browser control server status
openclaw browser status

# Start the browser control server
openclaw browser start

# Stop the browser
openclaw browser stop

# Reset the browser profile (clears cookies, cache, and state)
openclaw browser reset-profile

Tab Management

# List all open browser tabs
openclaw browser tabs

# Open a URL in a new tab
openclaw browser open --url https://docs.openclaw.ai

# Focus on a specific tab
openclaw browser focus --target-id e12

# Close a tab
openclaw browser close --target-id e12

DOM Interaction

# Click an element by CSS selector
openclaw browser click "#submit-button"

# Type into an input field
openclaw browser type "#search-box" "OpenClaw CLI commands"

# Simulate key press
openclaw browser press "Enter"

# Run arbitrary JavaScript in the page context
openclaw browser evaluate "document.title"

# Export the current page as a PDF
openclaw browser pdf

For a broader look at how OpenClaw browser automation compares to n8n's HTTP request nodes and workflow automation, see our OpenClaw vs n8n comparison.

Cron Scheduler Commands

OpenClaw has a built-in cron scheduler that lets you automate agent tasks on a schedule — daily summaries, hourly checks, weekly reports — all triggered without any external cron daemon. Tasks are defined with natural language messages sent to the agent at the specified time.

View Scheduled Jobs

# List all scheduled cron jobs
openclaw cron list

# List all jobs including disabled ones
openclaw cron list --all

# Show cron job status and next run times
openclaw cron status

Add Cron Jobs

# One-time job at a specific UTC datetime
openclaw cron add \
  --name "morning-brief" \
  --at "2026-04-01T09:00:00Z" \
  --message "Summarize my emails and calendar events for today"

# Recurring job using millisecond interval (every hour)
openclaw cron add \
  --name "hourly-check" \
  --every 3600000 \
  --message "Check for any priority items or messages I missed"

# Recurring job using cron expression (every Monday at 9 AM)
openclaw cron add \
  --name "weekly-report" \
  --cron "0 9 * * 1" \
  --message "Generate a summary of last week's tasks and plan this week"

Manage Jobs

# Run a job immediately (ignores schedule, triggers now)
openclaw cron run --id morning-brief

# Enable a disabled job
openclaw cron enable --id morning-brief

# Disable a job (keeps it configured but inactive)
openclaw cron disable --id morning-brief

# View the run history for a job
openclaw cron runs --id morning-brief --limit 10

# Edit an existing job
openclaw cron edit --id morning-brief

# Remove a job permanently
openclaw cron rm --id morning-brief

Pro tip: Always run openclaw cron list before adding new jobs to avoid scheduling conflicts or duplicate jobs with similar names.

Message & Broadcast Commands

The message command group lets you send messages, poll reactions, and interact with channels directly from the CLI — useful for deployment notifications, alerts, and automated broadcasts.

Send Messages

# Send a message to a Telegram user or group
openclaw message send \
  --channel telegram \
  --target @username \
  --message "Hello from the terminal!"

# Send to a Discord channel
openclaw message send \
  --channel discord \
  --target channel:123456789 \
  --message "Deployment successful ✓"

# Broadcast to all connected channels simultaneously
openclaw message broadcast --message "System maintenance in 30 minutes"

Polls & Reactions

# Create a poll in a Discord channel
openclaw message poll \
  --channel discord \
  --target channel:123 \
  --poll-question "Snack preference for the team meeting?" \
  --poll-option "Pizza" \
  --poll-option "Sushi" \
  --poll-option "Tacos"

# Add a reaction to a message
openclaw message react --channel telegram --message-id 789 --emoji "👍"

# List reactions on a message
openclaw message reactions --channel discord --message-id 456

Message Management

# Edit a sent message
openclaw message edit --channel telegram --message-id 123 --content "Updated text"

# Delete a message
openclaw message delete --channel discord --message-id 789

# Pin a message
openclaw message pin --channel telegram --message-id 123

# Unpin a message
openclaw message unpin --channel telegram --message-id 123

# Search messages in a channel
openclaw message search --channel discord --query "deployment"

# Read unread messages
openclaw message read --channel telegram

Security & Audit Commands

Security is built into the OpenClaw CLI. These commands let you audit your configuration, rotate secrets, manage API keys, and ensure your installation is hardened for production use.

# Full security audit (read-only — shows issues but doesn't fix)
openclaw security audit

# Audit and auto-fix detected issues
openclaw security audit --fix

# List all stored secrets
openclaw secrets list

# Rotate secrets
openclaw secrets reload

# View the security configuration
openclaw security configure

# Apply security rules from config
openclaw security apply

# Reset security settings to defaults
openclaw security reset

Run openclaw security audit weekly as a maintenance practice. It catches expiring API keys, authentication issues, misconfigured channel tokens, and provider quota problems before they cause outages.

Logs & Status Commands

These are your primary debugging and monitoring tools. Master these three and you can diagnose most OpenClaw issues in under a minute.

Logs

# View recent gateway logs
openclaw logs

# Stream logs in real time (tail -f style)
openclaw logs --follow

# Get last 200 log lines as JSON
openclaw logs --json --limit 200

# Filter logs by severity
openclaw logs --log-level error

# Filter logs to a specific channel
openclaw channels logs --channel telegram --follow

Status

# Full system status overview
openclaw status

# Deep status with all connected services
openclaw status --all --deep

# Provider usage and quota status
openclaw status --usage

# Status for a specific channel
openclaw channels status --probe

Doctor — The Most Useful Command in OpenClaw

# Run the health check (checks Node.js, gateway, config, auth, channels)
openclaw doctor

# Deep scan with verbose output
openclaw doctor --deep

# Auto-fix all detected issues
openclaw doctor --yes

# Combined deep scan + auto-fix
openclaw doctor --deep --yes

# Generate a new gateway token
openclaw doctor --generate-gateway-token

The doctor command checks: Node.js version compatibility, gateway connectivity, config file validity, provider authentication, channel status, and plugin health. Run it immediately after any upgrade or unexpected behavior.

Hooks Commands

Hooks are internal event listeners that extend OpenClaw's behavior without modifying the core gateway. They fire on specific lifecycle events — like agent startup, new messages, or gateway restart — making them powerful for custom automation.

# List all discovered hooks
openclaw hooks list

# Enable a specific hook
openclaw hooks enable session-memory

# Disable a hook
openclaw hooks disable command-logger

# Get detailed info about a hook
openclaw hooks info boot-md

Common Built-in Hooks

OpenClaw Built-in Hooks Reference
Hook Name	Trigger Event	Purpose
`session-memory`	`command:new`	Saves session context to memory when a new session starts
`command-logger`	`command`	Logs all CLI commands to an audit file
`boot-md`	`gateway:startup`	Runs the `BOOT.md` workspace file when the gateway starts
`soul-evil`	`agent:bootstrap`	Swaps the `SOUL.md` persona file during the purge window

System & Utility Commands

Dashboard

# Open the web dashboard in your default browser
openclaw dashboard

# Print the dashboard URL without opening it
openclaw dashboard --url-only

# Default URL: http://localhost:18789
# On headless/VPS machines: prints an SSH tunnel command instead

TUI (Terminal UI)

# Open the terminal-based interactive UI (no browser needed)
openclaw tui

Direct Agent Interaction

# Send a one-shot message to your agent from the CLI
openclaw agent --message "Summarize my tasks for today"

# Run one agent turn via the Gateway
openclaw gateway call --message "What is the weather in Delhi today?"

Update

# Update OpenClaw to the latest version
openclaw update

# Check available update channels
openclaw update channels list

Presence & Heartbeat

# View heartbeat status (agent online/offline detection)
openclaw system heartbeat last

# Enable heartbeat monitoring
openclaw system heartbeat enable

# Disable heartbeat
openclaw system heartbeat disable

# System-level event log
openclaw system event

# Presence status across connected channels
openclaw presence

Users & Permissions

# Add a team member with a role
openclaw users add --name "Alice" --role "developer"

# Set permissions for a user on a specific skill
openclaw permissions set --user "Alice" --skill "use-database" --allow

# List permissions for a user
openclaw permissions list --user Bob

# Filter activity logs by user
openclaw logs | grep "User: @Alice"

MCP & ACP Commands

MCP (Model Context Protocol) and ACP (Agent Client Protocol) are OpenClaw's integration layers for connecting external tools and coordinating multi-agent workflows.

MCP — Tool Integration

# Check MCP server status
openclaw mcp status

# View MCP health
openclaw mcp health

# Probe MCP connections
openclaw mcp status probe

# Discover available MCP servers
openclaw mcp discover

# Install an MCP server
openclaw mcp install my-mcp-server

# Uninstall an MCP server
openclaw mcp uninstall my-mcp-server

# Start an MCP server
openclaw mcp start my-mcp-server

# Stop an MCP server
openclaw mcp stop my-mcp-server

ACP — Multi-Agent Protocol

# Check ACP status (multi-agent coordination)
openclaw acp status

# View ACP health
openclaw acp health

ACP allows OpenClaw to spawn sub-agents — for example, delegating a coding task to a Claude Code sub-agent while the main agent handles communication and coordination. This is how OpenClaw handles complex, parallelizable workflows in 2026.

For a practical look at setting up Claude Code alongside OpenClaw, see our OpenClaw + Claude Code setup guide. For comparing OpenClaw to GitHub Copilot in coding workflows, see Claude Code vs GitHub Copilot.

Common Troubleshooting Sequence

When something goes wrong with OpenClaw, follow this diagnostic sequence before reaching for the nuclear reset option:

Step 1: Check gateway status

openclaw gateway status --deep --json

Step 2: Run doctor with auto-fix

openclaw doctor --deep --yes

Step 3: Tail logs for errors

openclaw logs --follow --log-level error

Step 4: Restart the gateway

openclaw gateway restart

Step 5: Check channel health

openclaw channels status --probe

Step 6: Validate config file

openclaw config file validate

Step 7: If all else fails — soft reset

openclaw reset --scope config+creds+sessions
openclaw onboard --install-daemon

Seeing a 1008 pairing error or disconnection issue with your browser extension? See our dedicated fix guide: OpenClaw 1008 disconnected / pairing required error fix.

Frequently Asked Questions

What is the single most important OpenClaw CLI command to know?: openclaw doctor is the most useful command in the entire CLI. Run it after any config change, after every upgrade, and whenever the gateway behaves unexpectedly. It checks Node.js version, gateway connectivity, config validity, provider authentication, and channel status — and can auto-fix most issues with --yes.
How do I start the OpenClaw gateway?: Use openclaw gateway start to start the background daemon, or openclaw gateway run to run it in the foreground for debugging. Check status at any time with openclaw gateway status.
How do I connect WhatsApp or Telegram to OpenClaw?: For Telegram: openclaw channels add --channel telegram --token YOUR_BOT_TOKEN. For WhatsApp: openclaw channels add whatsapp, then scan the QR code with your phone. Check out the full WhatsApp setup guide for step-by-step instructions.
Can I run OpenClaw without a messaging platform?: Yes. Use openclaw tui for a terminal-based interactive UI, or openclaw agent --message "your prompt" for direct one-shot interactions from the CLI.
How do I reset OpenClaw to factory defaults?: There are three levels: openclaw reset --scope config (config only), openclaw reset --scope config+creds+sessions (preserves workspace), and openclaw reset --scope full (nuclear — removes everything). Always try the lighter scopes first.
What is the difference between openclaw gateway run and openclaw gateway start?: gateway run launches the gateway in the foreground — blocks your terminal and shows live logs. Use this for debugging. gateway start installs and starts the gateway as a background system service (daemon) — ideal for production and always-on deployments.
How do I use multiple profiles or separate work and personal agents?: Use the --profile global flag: openclaw --profile work channels list. This isolates all state, config, and credentials under a separate directory. Alternatively, use the openclaw agents add command to create fully isolated agent instances with their own workspaces.

Conclusion

The OpenClaw CLI is one of the most comprehensive command-line interfaces in the AI agent space — covering everything from gateway management and channel setup to browser automation, multi-agent coordination, cron scheduling, and security auditing. Mastering it transforms OpenClaw from a chatbot into a true local AI operating system.

Here is a quick recap of the most-used command categories:

Daily drivers: openclaw doctor, openclaw gateway status, openclaw logs --follow
Setup & configuration: openclaw onboard, openclaw config get/set, openclaw channels add
Automation: openclaw cron add, openclaw browser open, openclaw message broadcast
Advanced: openclaw agents add, openclaw mcp install, openclaw acp status

Bookmark this page or save it as a PDF for offline reference — the most useful time to have a CLI cheatsheet is when the gateway is down and you need to debug without a working browser.

Continue exploring the Cloudvyn blog for more OpenClaw guides and AI developer content:

What Is the OpenClaw CLI?

Global Flags and Options

Setup & Onboarding Commands

First-Time Setup

Reset Commands

Uninstall

Gateway Commands

Start, Stop, Restart

Daemon Service Management

Gateway Sessions

Health Probe

Config Commands

Read Config Values

Write Config Values

Config File Operations

Backup & Restore

Channels Commands

View Channels

Add Channels

Remove Channels

Models Commands

List & Status

Set Active Model

Model Aliases

Agent Model Image

Agents Commands

List & Inspect

Add & Delete Agents

Bindings — Route Channels to Agents

Agent Identity

ACP — Agent Client Protocol

Skills Commands

List & Inspect

Directory

Writing a Custom Skill

Plugins Commands

Memory Commands

Browser Automation Commands

Browser Lifecycle

Tab Management

DOM Interaction

Cron Scheduler Commands

View Scheduled Jobs

Add Cron Jobs

Manage Jobs

Message & Broadcast Commands

Send Messages

Polls & Reactions

Message Management

Security & Audit Commands

Logs & Status Commands

Logs

Status

Doctor — The Most Useful Command in OpenClaw

Hooks Commands

Common Built-in Hooks

System & Utility Commands

Dashboard

TUI (Terminal UI)

Direct Agent Interaction

Update

Presence & Heartbeat

Users & Permissions

MCP & ACP Commands

MCP — Tool Integration

ACP — Multi-Agent Protocol

Common Troubleshooting Sequence

Step 1: Check gateway status

Step 2: Run doctor with auto-fix

Step 3: Tail logs for errors

Step 4: Restart the gateway

Step 5: Check channel health

Step 6: Validate config file

Step 7: If all else fails — soft reset

Frequently Asked Questions

Conclusion