OpenClaw Integration Guide

Target audience: Developers, Deployment Administrators Last updated: 2026-08-06

OpenClaw manages model Providers, Agents, and messaging channels simultaneously. When troubleshooting, separate these three layers: first confirm the model works in local conversation, then enable the gateway, and finally add messaging channels one at a time.

Installation and Initialization

The current version of OpenClaw requires Node.js 22 or later. You can use the install script:

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

Or install via npm:

npm install -g openclaw@latest

PowerShell:

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

After installation, run:

openclaw --version
openclaw onboard

You can skip messaging channel configuration during initialization. Complete the model Provider first — subsequent issues will be easier to diagnose.

Register a Provider

Edit ~/.openclaw/openclaw.json and merge the following fields into your existing configuration. Do not overwrite other Agent or channel settings in the file.

{
  "models": {
    "mode": "merge",
    "providers": {
      "myplatform": {
        "baseUrl": "https://<your-api-endpoint>",
        "apiKey": "${CUSTOM_API_KEY}",
        "api": "anthropic-messages",
        "models": [
          {
            "id": "deepseek-v4-flash",
            "name": "DeepSeek V4 Flash"
          }
        ]
      }
    }
  },
  "agents": {
    "defaults": {
      "model": {
        "primary": "myplatform/deepseek-v4-flash"
      }
    }
  }
}

Set the environment variable before starting:

export CUSTOM_API_KEY="YOUR_API_KEY"

This example uses Messages Compatible, so the Base URL does not include /v1. Whether the Messages protocol (/v1/messages) is available depends on the platform of the group bound to the Key and whether that group allows Messages dispatching. If unavailable, the platform returns 403 — switch to OpenAI Compatible instead. The reasoning, input, contextWindow, and maxTokens fields in OpenClaw custom models are optional capability declarations. Only fill them in after confirming the values from the current model information in the console. Do not copy them from other model configurations.

Use the endpoint address shown on the "API Key" page in the console.

Restart and check the status:

openclaw gateway restart
openclaw models list
openclaw status

The model list should show myplatform/deepseek-v4-flash. Then perform a local text Q&A session and verify the request in the console usage records.

Gateway Security

Keep the gateway authentication settings generated during OpenClaw initialization. If you need access from other devices, configure a Token following the current version's security guide. Do not set the remote gateway to no-authentication for convenience.

Run the following command to check and fix common security configurations:

openclaw doctor --fix

Even when the gateway is used only locally, you should restrict the listening address to avoid accidental exposure to the LAN or the public internet.

Feishu Integration

The current version of OpenClaw provides a Feishu channel login wizard. First create an enterprise custom app on the Feishu Open Platform, prepare the App ID and App Secret, then run:

openclaw channels login --channel feishu

The wizard will install the Feishu plugin if missing and guide you through entering credentials. On the app side, you need to enable the bot, configure message-related permissions, and subscribe to im.message.receive_v1. After completing the configuration:

openclaw gateway restart
openclaw channels status --probe

If a pairing code appears on the first direct message, an administrator can approve it:

openclaw pairing list feishu
openclaw pairing approve feishu PAIRING_CODE

Grant permissions based on your actual use case. Do not enable unnecessary document, drive, or contact permissions.

WeChat Integration

The WeChat channel is provided by an external plugin published by Tencent. Quick installation:

npx -y @tencent-weixin/openclaw-weixin-cli install

Or install and enable manually:

openclaw plugins install "@tencent-weixin/openclaw-weixin"
openclaw config set plugins.entries.openclaw-weixin.enabled true
openclaw gateway restart

Initiate QR code login:

openclaw channels login --channel openclaw-weixin

After logging in, verify:

openclaw plugins list
openclaw channels status --probe

The plugin package and OpenClaw version must be compatible with each other. If the installation succeeds but the channel fails to start, check the versions of both first. Do not modify a Provider that has already been verified.

DingTalk or QQ Integration

DingTalk and QQ are typically connected through external channel plugins. Package names, fields, and maintenance status may change. Follow this process:

1. Confirm the package name and minimum version in the channel directory or target plugin documentation for your current OpenClaw version.
2. Check the plugin publisher and latest update date.
3. Create an app or bot on the corresponding open platform, and only enable the permissions you need.
4. Install using openclaw plugins install <plugin-package>.
5. Preferably run openclaw channels add or the login wizard provided by the plugin.
6. After restarting, check with openclaw channels status --probe.

Do not directly copy channel JSON from other versions. When an external plugin requires you to save an App Secret, use the Secret or environment variable mechanism supported by your current version of OpenClaw.

Layered Troubleshooting

Test each channel individually after adding it. If you modify multiple channels at once before troubleshooting, it becomes difficult to determine whether the issue originates from the model, the gateway, or a plugin.