OpenClaw Usage Guide

OpenClaw is an open-source personal AI assistant platform that supports interaction via the command line, a web interface, and messaging channels such as DingTalk. This guide explains how to connect OpenClaw to OPEAI Platform to have conversations using Claude models.

Model Support

When connecting through OPEAI Platform, the following model is recommended:

Model	Description
claude-sonnet-4.6	Recommended — strong overall capability, fast response

For the full list of available models, log in to the OPEAI Platform Console.

Install OpenClaw

Requirements

• Node.js 22 or higher

macOS / Linux

Using the official install script:

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

Or install globally via npm:

npm install -g openclaw@latest

Windows

Using PowerShell:

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

Or install globally via npm:

npm install -g openclaw@latest

Verify Installation

openclaw --version

If the version number is printed, the installation was successful.

Complete the Onboarding Wizard

After the first installation, OpenClaw automatically launches a configuration wizard to help you complete the initial setup. You can also run openclaw onboard manually at any time.

Prompt	Recommended choice
I understand this is personal-by-default and shared/multi-user use requires lock-down. Continue?	Select Yes
Onboarding mode	Select QuickStart
Model/auth provider	Select Skip for now (configure OPEAI Platform later)
Filter models by provider	Select All providers
Default model	Select Keep current
Select channel (QuickStart)	Select Skip for now (configure channel later)
Configure skills now? (recommended)	Select No
Enable hooks?	Press Space to select an option, then press Enter to proceed
How do you want to hatch your bot?	Select Do this later

Configure OpenClaw (pointing to OPEAI Platform)

Connecting to OPEAI Platform requires the following key parameters:

Parameter	Value
baseUrl	https://api-platform.ope.ai/v1
apiKey	Obtain from the OPEAI Platform Console
api	openai-completions
models	Model list, e.g. claude-sonnet-4.6

Important: Make sure the reasoning parameter is set to false in the model configuration, otherwise responses may be empty.

Open ~/.openclaw/openclaw.json in a text editor and add the following content:

{
  "models": {
    "providers": {
      "platform": {
        "baseUrl": "https://api-platform.ope.ai/v1",
        "apiKey": "sk-xxxx",
        "api": "openai-completions",
        "models": [
          {
            "id": "claude-sonnet-4.6",
            "name": "Claude Sonnet 4.6",
            "reasoning": false
          }
        ]
      }
    }
  }
}

Note: Use your own Platform API Key and avoid exposing real secrets in screenshots, shell history, or logs.

Using OpenClaw

CLI Mode

openclaw tui

Launches an interactive terminal interface so you can chat with the AI directly in your terminal.

Web Dashboard

openclaw dashboard

Visit http://127.0.0.1:18789 to chat through your browser.

Chat Tool Integration

OpenClaw also supports integration with messaging tools such as DingTalk. Refer to the OpenClaw official documentation for configuration details.

Troubleshooting

Empty Replies

• Check that the reasoning parameter is set to false
• Confirm the Model ID is spelled correctly (e.g. claude-sonnet-4.6)
• Confirm the provider name matches the configuration

Model Not Found

• Confirm the model name is spelled correctly
• Log in to the OPEAI Platform Console to verify the model is available

command not found

• OpenClaw was not installed successfully. Run openclaw --version in your terminal; if a version number is printed, the installation succeeded
• Confirm Node.js version is ≥ 22
• Try reinstalling: npm install -g openclaw@latest