← Back to Skills Marketplace
lumint

Chrome Mcp

by lumint · GitHub ↗ · v0.1.0 · MIT-0
cross-platform ⚠ suspicious
875
Downloads
1
Stars
0
Active Installs
1
Versions
Install in OpenClaw
/install chrome-mcp
Description
通过 Chrome DevTools MCP 控制本机 Chrome 浏览器(已登录的真实会话)。 适用场景: - 浏览、阅读任意网页内容 - 操作 X (Twitter):浏览 feed、发推文、转帖、点赞、删帖 - 操作任何需要登录的网站(保留已有登录状态) - 截图、读取页面结构、执行 JS - "帮我看看...
README (SKILL.md)

Chrome MCP — 浏览器控制 Skill

通过 Chrome DevTools MCP 直接连接已运行的 Chrome,操控真实已登录的浏览器会话。

前提条件

Chrome 必须开启远程调试:打开 chrome://inspect/#remote-debugging 并启用开关。

工具优先级

优先使用 Chrome DevTools MCP 工具chrome__*),效率最高:

  • 无需截图即可读取页面结构(节省 ~90% token)
  • 直接操作 DOM 元素,稳定可靠
  • 连接已登录的 Chrome 会话

如果 MCP 工具不可用,回退到 ~/.claude/skills/chrome-cdp/scripts/cdp.mjs

Chrome DevTools MCP 核心工具

chrome__navigate          导航到 URL,等待页面加载完成
chrome__screenshot        截图(仅在需要视觉确认时使用)
chrome__get_accessibility_tree  获取页面结构(首选,比截图高效 10x)
chrome__click             点击元素(CSS 选择器)
chrome__type              在当前焦点输入文字
chrome__evaluate          执行 JavaScript
chrome__wait_for_element  等待元素出现

X (Twitter) 操作指南

浏览 Feed

1. chrome__navigate → https://x.com/home
2. chrome__get_accessibility_tree → 读取推文列表
3. 如需滚动加载更多:chrome__evaluate → window.scrollBy(0, 800)
4. 重复步骤 2-3 直到获取足够内容

发推文

1. chrome__navigate → https://x.com/home
2. chrome__click → [data-testid="tweetTextarea_0"]
3. chrome__type → 推文内容
4. chrome__click → [data-testid="tweetButtonInline"]

删除推文/取消转帖

删除自己的推文:

1. 定位到推文,chrome__click → [data-testid="caret"](三点菜单)
2. chrome__wait_for_element → [role="menuitem"]
3. chrome__click → 第一个 menuitem("删除")
4. chrome__click → [role="alertdialog"] [role="button"](确认)

取消转帖:

1. chrome__click → [data-testid="unretweet"]
2. chrome__click → [data-testid="unretweetConfirm"]

转帖 / 点赞

转帖: chrome__click → [data-testid="retweet"] 然后 [data-testid="retweetConfirm"]
点赞: chrome__click → [data-testid="like"]

通用操作技巧

  • 优先用 get_accessibility_tree 而非截图:结构清晰,消耗 token 极少
  • 等待加载:操作后如需等待,用 wait_for_element 而非 sleep
  • 长页面:通过 evaluate 执行 window.scrollBy 加载更多内容
  • 跨域 iframe 输入:用 type 而非 evaluate(JS eval 在跨域 iframe 中无法访问)
Usage Guidance
This skill appears internally consistent with its purpose, but it gives the agent potent access to your real browser sessions and lets it execute JavaScript inside pages. Before installing or using it, consider: - Only enable if you fully trust the skill and the agent that will invoke it. Treat trust carefully because the agent can read and act in any logged-in site. - Enable Chrome remote debugging only for a short period and only on a dedicated browser profile (create a separate Chrome profile with no sensitive logins). - Prefer manual, user-invoked use over autonomous invocation. If possible, disable autonomous invocation or require explicit user approval for every action. - Avoid using this skill while logged into sensitive accounts (banking, primary email, corporate SSO). - Limit actions to read-only where possible and avoid granting it permission to post or delete on social accounts unless you trust it. - Monitor network activity and running devtools ports; close the remote-debugging port when done. If you want a lower-risk setup, run Chrome in a disposable profile or a VM/container and keep all sensitive accounts out of that profile. If you need more assurance, request additional details from the skill author (how the chrome__* tools are provided/secured, whether there are safeguards to limit origins/actions, and why autonomous invocation is necessary).
Capability Analysis
Type: OpenClaw Skill Name: chrome-mcp Version: 0.1.0 The skill bundle provides tools for an AI agent to control a local, authenticated Chrome browser session, including capabilities for arbitrary JavaScript execution via 'chrome__evaluate' and instructions for destructive actions like deleting social media posts in SKILL.md. While these features are consistent with the stated purpose of browser automation, the ability to manipulate a user's live, logged-in web environment represents a high-risk capability that could be leveraged for unauthorized actions if the agent is misdirected. The reliance on a local fallback script (~/.claude/skills/chrome-cdp/scripts/cdp.mjs) also assumes a specific, potentially sensitive environment.
Capability Assessment
Purpose & Capability
Name/description align with the runtime instructions: the SKILL.md describes connecting to Chrome DevTools MCP and controlling a running, logged-in Chrome session. There are no unrelated required env vars, binaries, or installs.
Instruction Scope
The instructions explicitly direct the agent to read page structures (accessibility tree), take screenshots, execute arbitrary JavaScript (chrome__evaluate), and perform UI actions on any logged-in site (including X/Twitter). While this is consistent with the stated purpose, it grants the agent full access to the contents of any open, authenticated site (potentially including banking, email, etc.) and the ability to run JS inside pages — both capabilities can be used to exfiltrate sensitive data or perform unintended actions. The SKILL.md gives broad discretion (e.g., '操作任何需要登录的网站'), which increases risk.
Install Mechanism
Instruction-only skill with no install spec and no code files. This minimizes filesystem/install risk — nothing is downloaded or written by an installer as part of the skill bundle.
Credentials
The skill requests no environment variables, credentials, or config paths. The only external requirement is that Chrome be started with remote debugging enabled; that requirement is coherent with the stated functionality but has security implications (exposes a DevTools debugging endpoint).
Persistence & Privilege
always:false (not force-included), but model invocation is allowed (default). Combined with the skill's ability to access and control a logged-in browser and execute arbitrary JS, autonomous invocation increases blast radius. The skill does not request persistent system changes itself, but its runtime privileges (remote DevTools access to logged-in sessions) are powerful.
How to Use
  1. Make sure OpenClaw is installed (local or Docker)
  2. Run the install command in chat: /install chrome-mcp
  3. After installation, invoke the skill by name or use /chrome-mcp
  4. Provide required inputs per the skill's parameter spec and get structured output
Version History
v0.1.0
Initial release of chrome-mcp: - Enables direct control of a logged-in local Chrome session through Chrome DevTools MCP. - Supports browsing and interacting with any website, including X (Twitter): reading feeds, posting, retweeting, liking, and deleting tweets. - Optimizes operations by directly accessing the browser's DOM and accessibility tree for efficient content extraction. - Provides command examples for key actions (navigation, screenshot, JS execution, element click/type, waiting for elements). - Includes guidance for common use cases and best practices (e.g., avoiding screenshots when possible, using waits appropriately). - Requires Chrome to be started with remote debugging enabled.
Metadata
Slug chrome-mcp
Version 0.1.0
License MIT-0
All-time Installs 0
Active Installs 0
Total Versions 1
Frequently Asked Questions

What is Chrome Mcp?

通过 Chrome DevTools MCP 控制本机 Chrome 浏览器(已登录的真实会话)。 适用场景: - 浏览、阅读任意网页内容 - 操作 X (Twitter):浏览 feed、发推文、转帖、点赞、删帖 - 操作任何需要登录的网站(保留已有登录状态) - 截图、读取页面结构、执行 JS - "帮我看看... It is an AI Agent Skill for Claude Code / OpenClaw, with 875 downloads so far.

How do I install Chrome Mcp?

Run "/install chrome-mcp" in the OpenClaw or Claude Code chat to install it in one step — no extra setup required.

Is Chrome Mcp free?

Yes, Chrome Mcp is completely free, licensed under MIT-0. You can download, install and use it at no cost.

Which platforms does Chrome Mcp support?

Chrome Mcp is cross-platform and runs anywhere OpenClaw / Claude Code is available (cross-platform).

Who created Chrome Mcp?

It is built and maintained by lumint (@lumint); the current version is v0.1.0.

💬 Comments