Openclaw Quickstart

Install with one command.

macOS/Linux:

curl -fsSL https://openclaw.ai/install.sh | bash

Windows (PowerShell):

iwr -useb https://openclaw.ai/install.ps1 | iex

Onboarding

After installation, run Onboarding to complete initial configuration. This is the most critical step:

openclaw onboard --install-daemon

The onboarding wizard will configure authentication, Gateway settings, and optional channels:

Authentication Configuration: Set Gateway Token or Password to protect your API endpoints
Model Provider Configuration: Select and configure your LLM (e.g., Anthropic API Key)
Channel Configuration (Optional): Connect Telegram Bot, WhatsApp, and other chat channels
Install System Service: The --install-daemon parameter enables Gateway auto-start on boot

Gateway Basic Operations

Gateway is the core process of OpenClaw; all features depend on it running:

# View Gateway status
openclaw gateway status

# Run in foreground (good for debugging)
openclaw gateway --port 18789

# Verbose logging mode
openclaw gateway --port 18789 --verbose

# Force port occupation and start (resolve port conflicts)
openclaw gateway --force

# Restart Gateway
openclaw gateway restart

# Stop Gateway
openclaw gateway stop

# View real-time logs
openclaw logs --follow

Access Control UI

Control UI is OpenClaw's visual management interface:

# Open Control UI (automatically opens in browser)
openclaw dashboard

Or access directly at: http://127.0.0.1:18789/.

Main features of Control UI:

Chat Interface: Chat with AI directly in the browser without configuring any channels
Status Monitoring: View Gateway health status and channel connection status
Session Management: View and manage conversation history across different channels

Next, we can start chatting on the page:

🚀 Fastest Experience Path

Open Control UI and start chatting directly—no need to configure any Channel! This is the fastest way to verify OpenClaw is working properly.

Key Commands Quick Reference

Command	Description
`openclaw doctor`	Comprehensive health check and attempt to fix configuration issues
`openclaw health`	Quick health check
`openclaw status`	View overall Gateway and channel status
`openclaw gateway status`	View Gateway process status
`openclaw gateway status --deep`	Deep status check (includes RPC probes)
`openclaw gateway run`	Run Gateway in foreground
`openclaw gateway start`	Start Gateway service in background
`openclaw gateway stop`	Stop Gateway service
`openclaw gateway restart`	Restart Gateway service
`openclaw gateway probe`	Check Gateway connectivity and status
`openclaw gateway discover`	Discover local network or remote Gateway
`openclaw gateway call health`	Call Gateway health check interface
`openclaw channels status --probe`	Check all channel connection status
`openclaw dashboard`	Open browser control panel
`openclaw logs --follow`	Follow log output in real-time
`openclaw onboard`	Re-run initialization configuration wizard
`openclaw secrets reload`	Hot reload Secrets configuration

Gateway Hot Reload Mechanism

OpenClaw supports configuration hot reload, defaulting to hybrid mode:

`gateway.reload.mode`	Behavior Description
`off`	No automatic reload, manual restart required
`hot`	Apply only safe changes (does not interrupt connections)
`restart`	Restart Gateway for any change
`hybrid` (default)	Hot apply what can be hot applied, auto restart what needs restart

TUI: Terminal Chat Interface

In addition to the browser Control UI, OpenClaw also provides a fully terminal-based interactive interface—TUI (Terminal UI). No browser needed; you can chat directly even when SSH'd into a server.

Quick Start

# Step 1: Ensure Gateway is running
openclaw gateway

# Step 2: Open TUI
openclaw tui

Connect to remote Gateway:

openclaw tui --url ws://<host>:<port> --token <gateway-token>

# If Gateway uses password authentication
openclaw tui --url ws://<host>:<port> --password <password>

Interface Layout

After TUI starts, you will see four areas:

Area	Content
Header	Connection URL, current Agent, current Session
Chat log	User messages, AI responses, system notifications, tool call cards
Status line	Connection / running status (connecting / running / streaming / idle / error)
Footer	Connection status + Agent + Session + Model + Token count

Header area:

Footer area:

Keyboard Shortcuts

Shortcut	Function
`Enter`	Send message
`Esc`	Abort current operation
`Ctrl+C`	Clear input (press twice to exit)
`Ctrl+D`	Exit TUI
`Ctrl+L`	Open model selector
`Ctrl+G`	Open agent selector
`Ctrl+P`	Open session selector
`Ctrl+O`	Toggle tool output expand/collapse
`Ctrl+T`	Toggle Thinking display (will reload history)

Common Slash Commands

Use / to bring up commands:

Common commands:

/help                          # View help
/status                        # View connection status
/agent <id>                    # Switch Agent
/session <key>                 # Switch Session
/model <provider/model>        # Switch model (e.g., /model anthropic/claude-sonnet-4-6)
/think <off|minimal|low|medium|high>  # Set thinking depth
/deliver <on|off>              # Enable/disable message delivery to Provider
/new                           # Reset current Session
/abort                         # Abort current operation
/exit                          # Exit TUI

Local Shell Commands

In the input box, start with ! to execute local shell commands directly:

!ls -la      # List current directory
!cat log.txt # View file contents

⚠️ TUI will prompt for confirmation on first use of ! in each session; if denied, ! will be unavailable for the rest of that session.

Startup Options

Parameter	Description	Default
`--url <url>`	Gateway WebSocket address	Read from config or `ws://127.0.0.1:<port>`
`--token <token>`	Gateway Token authentication	—
`--password <password>`	Gateway password authentication	—
`--session <key>`	Specify Session	`main`
`--deliver`	Enable message delivery on startup	Off by default
`--thinking <level>`	Override thinking level	—
`--history-limit <n>`	Number of history entries to load	200
`--timeout-ms <ms>`	Agent timeout	Read from agents.defaults.timeoutSeconds

YouTip