Browser Automation Skill

English | 中文

🚀 Overview

A powerful browser automation skill for Claude Code that enables web scraping, data extraction, and content capture with semi-automated login support. Built on Playwright for reliable, cross-browser automation.

✨ Key Features

• 🔐 Semi-Automated Login - Browser opens visibly, pauses for manual login, then continues automation
• 🌐 Universal Web Scraping - Extract content from any website (static or dynamic)
• 📸 Screenshot Capture - Full page or element-specific screenshots
• 📊 Multiple Output Formats - JSON, Markdown, CSV
• 🎯 CSS Selector Support - Precise data extraction with custom selectors
• ⚡ Dynamic Content Handling - Waits for JavaScript-loaded content

📦 Installation

Prerequisites

• Python 3.7 or higher
• Claude Code CLI or Desktop App
• Git (for installation)

Step 1: Install Playwright

```bash pip install playwright playwright install chromium ```

Step 2: Install the Skill

Option A: Install from GitHub

```bash

Clone the repository

git clone https://github.com/YOUR_USERNAME/browser-automation-skill.git

Copy to Claude Code skills directory

cp -r browser-automation-skill ~/.claude/skills/browser-automation ```

Option B: Manual Installation

1. Download this repository as ZIP
2. Extract to `~/.claude/skills/browser-automation`
3. Ensure the directory structure matches: ``` ~/.claude/skills/browser-automation/ ├── SKILL.md ├── scripts/ │ └── scrape.py └── evals/ └── evals.json ```

Step 3: Verify Installation

Restart Claude Code and check if the skill appears:

```bash

The skill should now be available in Claude Code

Try: "Help me scrape content from a website"

```

🎯 Usage

In Claude Code (Recommended)

Simply describe what you want to scrape in natural language:

``` "Scrape the title and first paragraph from Wikipedia's Python page"

"Extract the top 3 trending repositories from GitHub"

"Capture a screenshot of Hacker News homepage and extract the top 5 news items" ```

The skill will automatically trigger when you mention web scraping, content extraction, or browser automation.

Command Line Tool

Use the included `scrape.py` script directly:

```bash

Basic scraping

python ~/.claude/skills/browser-automation/scripts/scrape.py https://example.com

With manual login

python ~/.claude/skills/browser-automation/scripts/scrape.py https://example.com --login

Extract specific fields

python ~/.claude/skills/browser-automation/scripts/scrape.py https://example.com
--selectors title:h1 price:.price-tag description:.product-desc

Save as markdown

python ~/.claude/skills/browser-automation/scripts/scrape.py https://example.com
--format md --output result.md ```

🔐 Semi-Automated Login Workflow

This is the core feature that sets this skill apart:

1. Browser Opens - Chromium launches in visible mode (not headless)
2. Navigate to URL - Automatically goes to the target website
3. ⏸️ Pause for Login - Script pauses with message: "Please log in manually. Press Enter when done."
4. 👤 You Log In - Complete login in the browser window (handle 2FA, CAPTCHA, etc.)
5. ▶️ Continue - Press Enter in terminal to resume automation
6. 📊 Extract Data - Script continues with data extraction
7. ✅ Complete - Returns structured results

Why Semi-Automated?

• Handles Complex Auth - Works with 2FA, CAPTCHA, OAuth
• No Credential Storage - Never stores passwords or tokens
• Human-in-the-Loop - You maintain control over authentication
• Reliable - Bypasses anti-bot detection that blocks headless browsers

🛠️ Dependencies

Required

• Python 3.7+ - Runtime environment
• Playwright - Browser automation library
• Chromium - Browser engine (installed via Playwright)

Optional

• Claude Code - For natural language triggering
• Git - For easy installation and updates

📄 License

MIT License - see LICENSE file for details

🙏 Acknowledgments

• Built with Playwright
• Designed for Claude Code

---

🚀 概述

一个强大的Claude Code浏览器自动化skill，支持网页抓取、数据提取和内容捕获，具有半自动登录支持。基于Playwright构建，可靠的跨浏览器自动化。

✨ 核心特性

• 🔐 半自动登录 - 浏览器可见打开，暂停等待手动登录，然后继续自动化
• 🌐 通用网页抓取 - 从任何网站提取内容（静态或动态）
• 📸 截图捕获 - 整页或特定元素截图
• 📊 多种输出格式 - JSON、Markdown、CSV
• 🎯 CSS选择器支持 - 使用自定义选择器精确提取数据
• ⚡ 动态内容处理 - 等待JavaScript加载的内容

📦 安装

前置要求

• Python 3.7或更高版本
• Claude Code CLI或桌面应用
• Git（用于安装）

步骤1：安装Playwright

```bash pip install playwright playwright install chromium ```

步骤2：安装Skill

方式A：从GitHub安装

```bash

克隆仓库

git clone https://github.com/YOUR_USERNAME/browser-automation-skill.git

复制到Claude Code skills目录

cp -r browser-automation-skill ~/.claude/skills/browser-automation ```

方式B：手动安装

1. 下载此仓库为ZIP
2. 解压到 `~/.claude/skills/browser-automation`
3. 确保目录结构匹配

步骤3：验证安装

重启Claude Code并检查skill是否出现。

🎯 使用方法

在Claude Code中使用（推荐）

用自然语言描述你想抓取的内容：

``` "从Wikipedia的Python页面抓取标题和第一段"

"提取GitHub上前3个热门仓库"

"截图Hacker News首页并提取前5条新闻" ```

🔐 半自动登录工作流程

这是使此skill与众不同的核心功能：

1. 浏览器打开 - Chromium以可见模式启动
2. 导航到URL - 自动访问目标网站
3. ⏸️ 暂停登录 - 脚本暂停等待手动登录
4. 👤 你登录 - 在浏览器窗口中完成登录
5. ▶️ 继续 - 按Enter恢复自动化
6. 📊 提取数据 - 脚本继续数据提取
7. ✅ 完成 - 返回结构化结果

📄 许可证

MIT许可证 - 详见LICENSE文件

---

⭐ 如果这个skill对你有帮助，请给个Star！