surfai 0.8.2

Natural language browser automation

SurfAI

🏄 Natural Language Browser Automation

Say what you want, SurfAI does the browsing.

Quick Start

# Install
pip install surfai

# Configure your LLM (choose one)
surfai config --model kimi --key sk-your-kimi-key
surfai config --model deepseek --key sk-your-deepseek-key

# Start automating
surfai "Go to GitHub and find the most starred Python project"

Features

• 🎯 Zero Code: Just say what you want in natural language
• 🤖 Multi-LLM: Supports Kimi, DeepSeek, OpenAI, Gemini
• 🔒 Local First: Runs on your machine, data stays private
• 📸 Screenshots: Capture pages automatically
• 📊 Data Export: Save results to CSV, JSON, Excel

Supported Models

Model	Provider	Setup
Kimi	Moonshot	surfai config --model kimi --key sk-xxx
DeepSeek	DeepSeek	surfai config --model deepseek --key sk-xxx
GPT-4o	OpenAI	surfai config --model openai --key sk-xxx
Gemini	Google	surfai config --model gemini --key xxx

Usage Examples

Command Line

# Extract data
surfai "Go to example.com and get all product prices"

# Take screenshot
surfai "Open github.com/microsoft and take a screenshot"

# Complex task
surfai "Search for 'Python tutorials' on Google, open the first result, summarize the content"

Python API

import asyncio
from surfai import Browser

async def main():
    async with Browser() as browser:
        await browser.goto("https://github.com/openclaw/openclaw")
        stars = await browser.extract("stars count")
        print(f"Stars: {stars}")
        await browser.screenshot("openclaw.png")

asyncio.run(main())

Configuration

Configuration is stored in ~/.surfai/config.json:

{
  "default_model": "kimi",
  "models": {
    "kimi": {
      "api_key": "sk-xxx",
      "base_url": "https://api.moonshot.cn/v1"
    },
    "deepseek": {
      "api_key": "sk-xxx",
      "base_url": "https://api.deepseek.com/v1"
    }
  }
}

License

MIT License - see LICENSE file for details.