Skip to content

🤖 OpenClaw - Self-hosted AI Assistant Platform

This page explains how to configure OpenClaw to access models through this site's gateway, including installation, configuration, and the complete steps for integration with this site.

Project Overview

OpenClaw is an open-source, self-hosted personal AI assistant platform that connects messaging apps to AI agents running on your own hardware. Designed for developers and power users, it lets you own an autonomous AI assistant without surrendering data control.

OpenClaw is fully open source (MIT license). You can browse the source code, submit issues, or contribute on GitHub.

Core Features

Multi-channel Integration

  • Cross-platform coverage: Supports Feishu, Discord, Slack, Microsoft Teams, and other major instant messaging platforms
  • Unified gateway: Manage all channels through a single Gateway process
  • Voice support: Voice interaction on macOS/iOS/Android
  • Canvas interface: Renders interactive Canvas interfaces

Self-hosted & Data Security

  • Fully self-hosted: Runs on your own machine or server
  • Open and transparent: MIT open source license, code is fully transparent
  • Localized data: Context and skills stored on your local machine, not in the cloud

AI Agent Capabilities

  • Persistent operation: Supports running in the background with persistent memory
  • Scheduled tasks: Supports cron-based scheduled tasks
  • Session isolation: Sessions isolated by agent/workspace/sender
  • Multi-agent routing: Supports multiple agents working together
  • Tool calling: Native support for tool calls and code execution

1️⃣ Create a Token in the Console

  1. Open the console token page: https://api-cs-al.naci-tech.com/console/token
  2. Click "Create Token".

2️⃣ Install OpenClaw

Requirements:

  • Node.js 22 or higher
  • An AI model API key (i.e., a token from this site)
npm install -g openclaw@latest

After installation, run the onboarding wizard:

openclaw onboard

3️⃣ Configure OpenClaw for this site

Configuration File Location

OpenClaw's configuration file is located at ~/.openclaw/config.json. It can be generated by the onboarding wizard or edited manually.

Configuration Example

Below is a configuration example using this site as the model provider. Replace $API_KEY with the token you created in the console:

{
  "models": {
    "providers": {
      "agtcloud": {
        "baseUrl": "https://api-cs-al.naci-tech.com/v1",
        "apiKey": "$API_KEY",
        "auth": "api-key",
        "api": "openai-completions",
        "models": [
          {
            "id": "gpt-4o",
            "name": "gpt-4o",
            "api": "openai-completions",
            "input": ["text", "image"],
            "contextWindow": 128000,
            "maxTokens": 16384
          },
          {
            "id": "claude-3-5-sonnet-20241022",
            "name": "claude-3-5-sonnet",
            "api": "openai-completions",
            "input": ["text", "image"],
            "contextWindow": 200000,
            "maxTokens": 8192
          }
        ]
      }
    }
  },
  "agents": {
    "defaults": {
      "model": {
        "primary": "agtcloud/gpt-4o",
        "fallbacks": ["agtcloud/claude-3-5-sonnet-20241022"]
      }
    }
  }
}

Key Configuration Reference

Config Item Description
models.providers.agtcloud.baseUrl This site's gateway URL; must include /v1
models.providers.agtcloud.apiKey Your token from this site, created in the console
models.providers.agtcloud.models Model list; you can add any model supported by this site as needed
agents.defaults.model.primary Default primary model; format is provider/model-id
agents.defaults.model.fallbacks Fallback model list; automatically switches when the primary model is unavailable

Tip: Available model IDs can be viewed in this site's console under "Channel Management", or refer to the API Reference.

4️⃣ Start the Service

After configuration, start OpenClaw:

openclaw start

Once started, you can interact with the AI assistant through your configured channels (e.g., Feishu, Discord).

5️⃣ Verify this site's gateway usage

  1. Send a message to OpenClaw through one of your configured channels.
  2. Return to this site's console and confirm in "Usage Logs" that the corresponding call has been recorded and grouped as openclaw.

If you can see the call records in the console, OpenClaw is successfully accessing models through this site's gateway.