Quick Start Guide

Get your first AI-powered cloud desktop running in minutes. This guide will walk you through creating a basic cloud desktop using CodecFlow.

Prerequisites

• A CodecFlow account (sign up here)
• Your API key from the dashboard
• Basic knowledge of TypeScript/Python

1. Choose your preferred SDK:

   TypeScript

   npm install @codecflow/sdk

   yarn add @codecflow/sdk

   Python

   pip install codecflow-sdk

2. Initialize CodecFlow

   TypeScript

   import { CodecFlow } from '@codecflow/sdk';
   
   const codecflow = new CodecFlow({
     apiKey: 'your-api-key'
   });

   Python

   from codecflow import CodecFlow
   
   codecflow = CodecFlow(
       api_key='your-api-key'
   )

3. Create Your Cloud Desktop

   TypeScript

   const desktop = await codecflow.createDesktop({
     os: 'windows',  // or 'macos', 'linux'
     version: '11',  // OS version
     settings: {
       memory: '8gb',
       cpu: 4,
       secureMode: 'tee'  // Use Trusted Execution Environment
     }
   });

   Python

   desktop = await codecflow.create_desktop(
       os='windows',  # or 'macos', 'linux'
       version='11',  # OS version
       settings={
           'memory': '8gb',
           'cpu': 4,
           'secure_mode': 'tee'  # Use Trusted Execution Environment
       }
   )

4. Add Browser Control

   TypeScript

   const browser = await desktop.createBrowser();
   
   await browser.navigate('https://example.com');
   await browser.click('#some-button');
   await browser.type('#search-input', 'Hello World');

   Python

   browser = await desktop.create_browser()
   
   await browser.navigate('https://example.com')
   await browser.click('#some-button')
   await browser.type('#search-input', 'Hello World')

5. Connect Your AI Agent with MCP

   TypeScript

   const agent = await desktop.connectAgent({
     model: 'gpt-4',
     protocol: 'mcp',  // Model Context Protocol
     config: {
       prompt: 'You are an automation assistant...',
       capabilities: ['browser', 'filesystem', 'network']
     }
   });
   
   await agent.controlDesktop(desktop);

   Python

   agent = await desktop.connect_agent(
       model='gpt-4',
       protocol='mcp',  # Model Context Protocol
       config={
           'prompt': 'You are an automation assistant...',
           'capabilities': ['browser', 'filesystem', 'network']
       }
   )
   
   await agent.control_desktop(desktop)

6. Execute Tasks

   TypeScript

   await desktop.start();
   // Monitor desktop status
   desktop.on('status', (status) => console.log('Desktop status:', status));
   
   // Handle errors
   desktop.on('error', (error) => console.error('Desktop error:', error));
   
   // Shut down when done
   await desktop.shutdown();

   Python

   await desktop.start()
   # Monitor desktop status
   desktop.on('status', lambda status: print('Desktop status:', status))
   
   # Handle errors
   desktop.on('error', lambda error: print('Desktop error:', error))
   
   # Shut down when done
   await desktop.shutdown()