OpenClaw Configuration

1. Overview

OpenClaw (formerly Clawdbot) is an open-source AI agent framework that supports flexible configuration through configuration files or a GUI interface. The LongCat series models are supported for use within OpenClaw. This document explains how to integrate OpenClaw with the LongCat API Platform.

2. Prerequisites

Before using OpenClaw, you need to prepare the following:

OpenClaw Installation Package

Installation commands for macOS:

# Using npm
npm install -g openclaw@latest

# Or using pnpm
pnpm add -g openclaw@latest

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

Installation commands for Windows PowerShell:

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

LongCat Core Configuration

Confirm your API Keys (available on the API Keys).
Ensure that your account has sufficient token balance (check on the Usage Information).

Supported Models:

Model Name	API Format	Description
LongCat-Flash-Chat	OpenAI/Anthropic	High-performance general-purpose dialogue model
LongCat-Flash-Thinking	OpenAI/Anthropic	Deep thinking model
LongCat-Flash-Thinking-2601	OpenAI/Anthropic	Upgraded deep thinking model
LongCat-Flash-Lite	OpenAI/Anthropic	Efficient MoE model

3. Quick Start

Onboarding Command

openclaw onboard --install-daemon

Onboarding Configuration Options

Step	Recommendation
Risk Confirmation	Select yes to continue
Onboarding Mode	QuickStart is recommended for quick basic configuration
Model/Auth Provider	Select Skip for now
Filter models by provider	Select All providers
Default model	选择 Keep current
Select Channel	If no suitable option, select Skip for now and configure later
Configure Skills	Yes is recommended to enable local utility skills
Preferred node manager for skill installs	select npm
Install missing skill dependencies	select Skip for now
API Key Settings	Select No or Skip if not available
Gateway Service	Select Install or Reinstall to install Gateway service
How to Hatch Your Bot	Hatch in TUI is recommended for interactive experience

4. Post-Start Configuration

After completing the onboarding process, OpenClaw will automatically start the Gateway service and open the web control page.

Default Access Address

http://127.0.0.1:18789

If the page does not open automatically, you can manually access the above address in your browser.

Configuration File Location

~/.openclaw/openclaw.json

5. LongCat Model Configuration

After startup, you can configure LongCat models by modifying custom configurations.

Configuration Option 1: Modify Configuration File

1. Add Custom Model Provider

Add the models field to openclaw.json:

{
   "models": {
      "mode": "merge",
      "providers": {
         "longCat": {
            "baseUrl": "https://api.longcat.chat/openai",
            "apiKey": "YOUR_API_KEY",
            "api": "openai-completions",
            "authHeader": true,
            "models": [
               {
                  "id": "LongCat-Flash-Chat",
                  "name": "LongCat-Flash-Chat",
                  "reasoning": false,
                  "input": ["text"],
                  "contextWindow": 200000,
                  "maxTokens": 8192,
                  "compat": {
                     "maxTokensField": "max_tokens"
                  }
               }
            ]
         }
      }
   }
}

2. Modify Default Model Settings

Modify the agents field to set the default model:

{
   "agents": {
      "defaults": {
         "model": {
            "primary": "longCat/LongCat-Flash-Chat"
         }
      }
   }
}

Changes take effect immediately after saving.

Configuration Option 2: GUI Interface Configuration

In the web control page, go to Config → Models → Providers
Add the following configuration:

Configuration Item	Value
Api	`openai-completions`
Api Key	LongCat's API Key
Base Url	`https://api.longcat.chat/openai`
models - id	LongCat model name
models - name	LongCat model name

Complete Configuration Example

{
  "meta": {
    "lastTouchedVersion": "2026.1.24-3",
    "lastTouchedAt": "2026-01-29T02:21:28.058Z"
  },
  "wizard": {
    "lastRunAt": "2026-01-29T02:13:14.153Z",
    "lastRunVersion": "2026.1.24-3",
    "lastRunCommand": "onboard",
    "lastRunMode": "local"
  },
  "auth": {
    "profiles": {
      "anthropic:default": {
        "provider": "anthropic",
        "mode": "api_key"
      }
    }
  },
  "models": {
    "mode": "merge",
    "providers": {
      "longCat": {
        "baseUrl": "https://api.longcat.chat/openai",
        "apiKey": "YOUR_API Key",
        "auth": "api-key",
        "api": "openai-completions",
        "authHeader": true,
        "models": [
          {
            "id": "LongCat-Flash-Chat",
            "name": "LongCat-Flash-Chat",
            "reasoning": false,
            "input": ["text"],
            "contextWindow": 200000,
            "maxTokens": 8192,
            "compat": {
              "maxTokensField": "max_tokens"
            }
          }
        ]
      }
    }
  },
  "agents": {
    "defaults": {
      "model": {
        "primary": "longCat/LongCat-Flash-Chat"
      },
      "workspace": "/Users/xxx/.openclaw/workspace",
      "contextPruning": {
        "mode": "cache-ttl",
        "ttl": "1h"
      },
      "compaction": {
        "mode": "safeguard"
      },
      "heartbeat": {
        "every": "30m"
      },
      "maxConcurrent": 4,
      "subagents": {
        "maxConcurrent": 8
      }
    }
  },
  "messages": {
    "ackReactionScope": "group-mentions"
  },
  "commands": {
    "native": "auto",
    "nativeSkills": "auto"
  },
  "gateway": {
    "port": 18789,
    "mode": "local",
    "bind": "loopback",
    "auth": {
      "mode": "token",
      "token": ""
    },
    "tailscale": {
      "mode": "off",
      "resetOnExit": false
    }
  },
  "skills": {
    "install": {
      "nodeManager": "npm"
    }
  }
}

6. Getting Started

Once the configuration takes effect, you can start using OpenClaw.

Open TUI and check Gateway status

openclaw tui
/status

Open Web UI and interact on the Chat page

openclaw dashboard

Then enter a test message such as: "Hello, please introduce yourself"

If the configuration is correct, you will receive a response from the LongCat model.