Playwright Browser

This is a comprehensive Playwright wrapper exposing 101 browser automation tools through MCP. You get the full stack: launch browsers (Chromium, Firefox, WebKit), navigate pages, click elements, fill forms, extract content, take screenshots, intercept network requests, and monitor performance. The documentation is thoroughly bilingual (English and Chinese) with clear AI calling patterns, which is helpful if you're building agents that need to scrape data, test web apps, or automate repetitive browser tasks. It supports device emulation and video recording out of the box. The sheer number of tools means you can handle edge cases like frame switching, cookie manipulation, and accessibility tree queries without dropping down to custom code. If you need programmatic browser control and don't want to write Playwright bindings yourself, this covers most scenarios.

skills.sh installs: 135
GitHub stars: 71
last updated: Jun 3, 2026

View on GitHub

Install

$ npx -y skills add 91fapiao-cn/playwright-browser-skill --skill playwright-browser --agent claude-code

Installs into .claude/skills of the current project.

Files

Playwright Browser Skill - 浏览器自动化技能

⚡ AI 调用指南（OpenClaw 必读）

如何调用 MCP 工具

当用户请求浏览器操作时，你需要通过 MCP 协议调用相应的工具。以下是调用方法：

基本调用流程

启动浏览器 → 调用 browser_launch
访问网页 → 调用 browser_goto
执行操作 → 调用相应工具（点击、填写、提取等）
关闭浏览器 → 调用 browser_close

常用工具快速参考

用户需求	调用工具	参数示例
"访问网站"	`browser_goto`	`{ "url": "https://example.com" }`
"点击按钮"	`browser_click`	`{ "selector": "button.submit" }`
"填写表单"	`browser_fill`	`{ "selector": "#username", "value": "admin" }`
"获取标题"	`browser_get_title`	`{}`
"截图"	`browser_screenshot`	`{ "path": "screenshot.png", "fullPage": true }`
"获取文本"	`browser_get_text`	`{ "selector": "h1" }`
"等待元素"	`browser_wait_for_selector`	`{ "selector": ".content" }`

完整示例：访问网页并获取标题

用户："访问 example.com 并获取页面标题"

你应该调用：
1. browser_launch({ "headless": false })
2. browser_goto({ "url": "https://example.com" })
3. browser_get_title({})
4. browser_close({})