OpenClaw

OpenClaw supports three primary installation paths: script installer (fastest), global CLI installation, and building from source. This guide demonstrates installation using the script installer—the simplest official method—which automatically detects your OS, installs Node.js (if missing), and runs the onboarding wizard.

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

After running, it will:

1. • Install the OpenClaw CLI.
   • Prompt you for basic configuration (workspace, channels, etc.).
   • Automatically install the daemon (background service) to run 24/7.
   • If errors occur (e.g., permission issues): Use sudo or check if curl is installed.

After installation, the initial setup will run automatically. Simply follow the step-by-step instructions.

1. • It will verify the model provider and model name. Since configuration options for OpenAI-compatible APIs are not provided, you must first skip this step to complete the entire process before configuring the Provider and other details in the settings file. When using the Default model, select “Manually input model” and enter canopywave/moonshotai/kimi-k2.5 here.
   
   • channels: Discord/Telegram/WhatsApp, the Discord used in the tutorial.
   • Step 1: First, configure the Discord bot's permissions. If the corresponding Discord bot is already set up, simply enter the relevant information. If not, refer to Step 5 to configure the Discord bot.
   
   • After completion, you'll need to configure the bot's skills. Additional skills can be added later as needed.
   

After completion, press Ctrl+C to terminate the current openclaw process.

Then navigate to the configuration file to set our URL and key.

Open the file ~/.openclaw/openclaw.json using vim or nano. After performing the steps above, you'll see some content in the file. Locate the line preceding “agent” and add the following content to correctly configure the third-party API:

"models": {
    "mode":"merge",
    "providers": {
      "canopywave": {
        "baseUrl": "https://api.canopywave.io/v1",
        "apiKey": "your Canopy Wave API Key",
        "api": "openai-completions",
        "models": [
          {
            "id": "moonshotai/kimi-k2.5",
            "name": "Kimi K2.5",
            "reasoning": true,
            "contextWindow": 204800,
            "maxTokens": 8192
          }
        ]
      }
    }
  },

1. • After saving: You can use openclaw doctor to verify whether the configuration was successful. After configuration is complete, use openclaw onboard to reload. Once canopywave is added to the configuration file, you can skip the Provider settings and view the canopywave model within the model.