Browser Use
Browser agents and stealth cloud browsers for AI: run web tasks, fill forms, extract data.
作者提供的说明(英文原文)
Is this page readable for AI search? Crawler access, structure and citability checks, runs locally.
这三项永远不会合并成一个分数:意见、测量和人气是不同的东西。
我们还没有为它写通俗的介绍。下面是我们从注册中心读取到的事实。
选择你使用的 AI 应用,然后复制设置。
你需要
你用的是哪个应用?
在上方选择你的应用。我们会记住它,下次按钮会显示“安装到 …”。
打开 Claude Desktop → Settings → Developer → Edit Config,把下面的内容粘贴到 claude_desktop_config.json
{
"mcpServers": {
"geo-tool-check": {
"command": "npx",
"args": [
"-y",
"geo-tool-check",
"--mcp"
]
}
}
}在终端中运行这条命令
claude mcp add geo-tool-check -- npx -y geo-tool-check --mcp粘贴到项目中的 .cursor/mcp.json,或 ~/.cursor/mcp.json
{
"mcpServers": {
"geo-tool-check": {
"command": "npx",
"args": [
"-y",
"geo-tool-check",
"--mcp"
]
}
}
}浏览器会先询问是否打开应用,然后应用会显示设置并请你确认。密钥的值会留空,由你自己填写。
粘贴到项目中的 .vscode/mcp.json
{
"servers": {
"geo-tool-check": {
"type": "stdio",
"command": "npx",
"args": [
"-y",
"geo-tool-check",
"--mcp"
]
}
}
}浏览器会先询问是否打开应用,然后应用会显示设置并请你确认。密钥的值会留空,由你自己填写。
这个服务器在你的电脑上运行,而 ChatGPT 只能连接互联网上的服务器。请选择其他应用。
你的选择通过 Cookie 保存在这台设备上,无需账户。
这些设置是根据作者在注册中心的条目生成的。如果 README 的写法不同,请以 README 为准。
它在你的电脑上运行;安装后,工具列表会显示在你的 AI 应用中。
我们检查的是任何人都能看到的信息;我们没有阅读代码。不适用的检查项不计入总数。 · 检查于 2026年11月7日
注册名称已通过 GitHub 账户 shufflethis 验证,该账户拥有此仓库
我们还没有从 GitHub 读取这个仓库
我们还没有从 GitHub 读取这个仓库
它在你的电脑上运行,没有互联网端点
在你的电脑上运行,改用本地设置
它在你的电脑上运行;我们只能读取互联网端点的工具列表
与注册中心里这个分类中所有活跃的服务器比较。只有当发布者拥有该仓库时,星标数才计入。
它在 1 项适用的基础检查中通过了 1 项。我们没有阅读代码。在添加密钥或令牌之前,请先看看这个服务器要求哪些权限,并使用只具备所需权限的密钥。
注册中心没有列出必需的密钥,不过有些服务器会在首次使用时要求你登录。
它在你的电脑上运行。你的 AI 应用会在需要时启动它,数据不会经过作者的服务器(除非它自己调用在线服务)。
GitHub 仓库最后更新于–,最新版本(1.0.1)于2个月前进入注册中心。
支持 MCP 的应用,例如 Claude Desktop、Claude Code、Cursor 和 VS Code。你的电脑上还需要有 Node.js。
同一分类中有:Browser Use, Chrome DevTools MCP, MCP Server Browser
同一分类中其他作者的服务器,按星标数排序。
如何看卡片:★ GitHub 星标i三大支柱:评价 · 检查 · 使用情况i更新 = 最新版本发布到注册中心的时间i
Browser agents and stealth cloud browsers for AI: run web tasks, fill forms, extract data.
MCP server for Chrome DevTools
MCP server for browser use access
按原文从作者的仓库复制。PopNx MCP 没有审查或背书其内容。HTML、图片和脚本已被移除。链接会离开本站。
Is this page readable for AI search? One command answers it: crawler access, structure, and citability for ChatGPT, Perplexity, Claude, and Google AI — scored 0–100 with concrete findings.
npx geo-tool-check example.com
Everything runs on your machine. No account, no API key, no telemetry — nothing is ever sent to geo-tool.com. The score is computed by the same engine as the free website audit on geo-tool.com, so the number here matches the number there.
geo-tool-check — Node 18+, zero runtime dependenciesio.github.shufflethis/geo-tool-check (stdio server, ships in the package)https://www.geo-tool.com/mcp (Streamable HTTP) · manifest /.well-known/mcp.jsoncheck_citability scores a text passage with no network request at all--min-score 70 turns the score into an exit codeNot for measuring whether AI answers actually mention your brand — that is a different measurement: the hosted geo-tool.com workspace checks real ChatGPT, Perplexity, and Gemini answers for your buying-intent questions daily.
| I want to… | Run |
|---|---|
| Score a page | npx geo-tool-check example.com |
| See which AI crawlers robots.txt blocks | npx geo-tool-check example.com --crawlers |
| Gate a deploy in CI | npx geo-tool-check https://example.com --min-score 70 |
| Get machine-readable output | npx geo-tool-check example.com --json |
| Use it from an AI agent (MCP) | npx geo-tool-check --mcp — setup guide |
Almost every MCP client accepts this block (per-client setup, config files to copy):
{
"mcpServers": { "geo-tool-check": { "command": "npx", "args": ["-y", "geo-tool-check", "--mcp"] } } }
Clients that speak Streamable HTTP to remote servers can skip the install and use the hosted endpoint instead — same tools, rate-limited, fetches run through geo-tool.com's infrastructure rather than your machine:
claude mcp add --transport http geo-tool https://www.geo-tool.com/mcp
The three tools, identical in both variants:
| Tool | Arguments | What it answers |
|---|---|---|
check_ai_readiness | url, optional lang | Full 0–100 readiness score with per-category breakdown and findings |
check_ai_crawlers | url, optional lang | Which AI crawlers (GPTBot, OAI-SearchBot, ClaudeBot, …) robots.txt blocks |
check_citability | content, optional lang | How quotable a draft passage is — no network request at all |
Install once:
/plugin marketplace add geo-tool-com/geo-tool-check
/plugin install geo-tool-check@geo-tool
Then check any page with one command — works immediately, no restart needed:
/geo-tool-check example.com
How do I check whether ChatGPT can read my website? npx geo-tool-check example.com. Two things must be true: robots.txt must not block OpenAI's crawlers, and the content must exist in the plain HTML — ChatGPT's crawlers do not execute JavaScript.
Does robots.txt really affect AI visibility? It can end it. Disallow: / for GPTBot means OpenAI's systems never see your content. The checker flags blocked scored crawlers separately from training-only bots, because the consequences differ.
Why does a JavaScript-rendered page score poorly? Because AI crawlers see what curl sees. If content only exists after client-side rendering, models find an empty shell — the low score is the finding. Fixes: SSR, static generation, or prerendering.
Do I need an llms.txt file? Optional and cheap. Google states it does not use it; several AI tools read it. The check detects a real one and ignores SPA catch-all fakes.
Does geo-tool-check upload my content anywhere? No. Every fetch goes from your machine to the target site — never to geo-tool.com. The build fails if a bundle would try.
More: full FAQ.
Node runs no JavaScript — and neither do GPTBot, ClaudeBot, or PerplexityBot. A client-rendered page that scores low here is the finding, not a measurement gap. To compare the crawler's view against the rendered page, use the browser extension.
This checker measures whether a page can be read and cited. Whether AI systems actually mention your brand in real answers is a different measurement: the hosted geo-tool.com workspace checks real AI answers (ChatGPT, Perplexity, Gemini) for your buying-intent questions daily.
MIT. Built by track by track GmbH.