跳到正文
TakoAPI
返回技能列表

mcp-chrome

由 femto 提供

Browser & Automation
412下载量由 clawskills.sh 提供
2星标
1浏览量
0 个赞登录后即可为这个技能点赞
—暂无评分
你的评分

简介

Control Chrome browser with AI using MCP protocol.

描述

该技能的功能

Connects AI agents to your existing Chrome browser via a native bridge and extension. Supports clicking, form filling, screenshots, content extraction, history search, bookmark management, and network monitoring. Works with your active login sessions and browser settings.

Unlike Playwright-based tools that launch a fresh browser instance, this uses your already-running Chrome so existing sessions, extensions, and cookies are immediately available.

何时使用

  • Automating a multi-step login and form submission on a work portal
  • Taking screenshots of competitor pricing pages for comparison
  • Searching browsing history to find a page visited last week
  • Extracting product details from an e-commerce site without re-logging in
  • Monitoring network requests while debugging a web app

示例工作流

输入

User asks: 'Go to github.com and take a full-page screenshot'

智能体
  1. 1

    Calls chrome_navigate with url 'https://github.com'

  2. 2

    Waits for page to load

  3. 3

    Calls chrome_screenshot with fullPage set to true

  4. 4

    Returns the captured image to the user

输出

Full-page screenshot of github.com delivered in the conversation

前置要求

  • mcp-chrome-bridger npm package installed globally
  • Chrome extension loaded from GitHub Releases with Developer mode enabled

安装

clawhub install femto/mcp-chrome

运行 CLI 命令,或把这段提示词粘贴到 Claude Code、Cursor、Codex 或你在用的任意编程智能体里。

链接