• Published:

OpenClaw, formerly known as ClawdBot, has picked up a lot of attention lately, and after trying it out, it’s easy to see why. It’s a capable tool, and getting it running in a mainland China network environment is fairly straightforward once the basic pieces are in place. Here’s a practical walkthrough using a local Windows setup as the example, along with the steps for linking it to a QQ bot.

What you need first

Before deployment, prepare the following:

  • a model API
  • hardware with at least 2 CPU cores and 4 GB of RAM
  • Linux, macOS, or Windows
  • a QQ account

For the large model API, DeepSeek can be used as an example. Its official purchase page is:

https://platform.deepseek.com

The example below uses local deployment on Windows. The process is essentially the same on Linux and macOS; the main difference is the exact configuration path. OpenClaw stores its default config under the user directory on all three major systems.

Install the required environment

Start by installing Node.js. You need version v22.22.0 or newer:

https://nodejs.org/zh-cn/download

You’ll also need Git:

https://git-scm.com/install/windows

Allow PowerShell scripts to run

After installing the environment, system variables do not always refresh immediately. If necessary, restart the computer first. Then open PowerShell as an administrator and run the following command. When prompted, choose A to allow script execution.

Set-ExecutionPolicy -ExecutionPolicy RemoteSigned -Scope CurrentUser

Network-related setup for use in China

Set the npm registry mirror:

npm config set registry https://registry.npmmirror.com

Optionally, you can configure a global Git acceleration rule:

git config --global url."https://hub.cmoko.com/https://github.com/".insteadOf https://github.com/

If you want to remove that Git acceleration setting later, use:

git config --global --unset url."https://hub.cmoko.com/https://github.com/".insteadOf

Deploy OpenClaw

1. Install OpenClaw with npm

npm install -g openclaw-cn@latest

2. Initialize the configuration

openclaw-cn setup

3. Open the configuration directory

%USERPROFILE%\.openclaw

4. Edit openclaw.json

Replace the contents with the following configuration and fill in the relevant fields:

{
  "gateway": {
    "mode": "local",
    "bind": "loopback",
    "port": 18789,
    "auth": {
      "mode": "token",
      "token": "设置一个token令牌"
    }
  },

  "agents": {
    "defaults": {
      "model": {
        "primary": "openai-compat/填写模型名称"
      },
      "elevatedDefault": "full",
      "compaction": {
        "mode": "safeguard"
      },
      "maxConcurrent": 4
    }
  },

  "models": {
    "mode": "merge",
    "providers": {
      "openai-compat": {
        "baseUrl": "https://api.example.com/v1",
        "apiKey": "填写你的模型API密钥",
        "api": "openai-completions",
        "models": [
          {
            "id": "填写模型名称",
            "name": "填写模型名称"
          }
        ]
      }
    }
  },

  "tools": {
    "exec": {
      "backgroundMs": 10000,
      "timeoutSec": 1800,
      "cleanupMs": 1800000,
      "notifyOnExit": true
    },
    "elevated": {
      "enabled": true,
      "allowFrom": {}
    },
    "allow": [
      "exec",
      "process",
      "read",
      "write",
      "edit",
      "web_search",
      "web_fetch",
      "cron"
    ]
  },

  "channels": {}
}

A few fields matter most here:

  • token: set your own gateway token
  • primary: enter the model name you want to use
  • baseUrl: point this to your model provider’s API endpoint
  • apiKey: add your model API key
  • id and name: fill in the same model name configured with your provider

Create and connect the QQ bot

Next, create the bot itself. Open the QQ bot OpenClaw section in your browser, then create the bot directly there.

After that, install the bot plugin:

openclaw-cn channels add

Follow the interactive prompt, choose QQ (Community Edition), then enter the AppID and AppSecret when asked.

Start OpenClaw

Once configuration and bot integration are done, start the gateway:

openclaw-cn gateway start

At that point, the deployment is complete and you can start chatting with your QQ bot.

Useful commands

# 查看状态
openclaw-cn gateway status

# 重启
openclaw-cn gateway restart

# 查看日志
openclaw-cn logs

# 停止
openclaw-cn gateway stop

# 卸载
npm uninstall -g openclaw-cn

Notes and common issues

On Linux and macOS, the deployment process is the same in general; the only practical difference is that the configuration file lives under the user directory for that system.

If you want to access the web interface, include the token in the URL:

http://127.0.0.1:18789/?token=你的网关令牌

If something goes wrong, check the logs first:

openclaw-cn logs

Chinese documentation

https://clawd.org.cn/reference/cli-cheatsheet.html

Related Posts