Puppeteer Vision MCP Server

@djannot

代码仓库

基本信息

分类

浏览器自动化

传输方式

stdio

作者

djannot

提交者

Denis Jannot

关于 Puppeteer Vision MCP Server

暂无概览

#puppeteer #vision #openai

Puppeteer Vision MCP Server 配置

使用下面的配置,将此服务器添加到你的 MCP 客户端。

{
  "mcpServers": {
    "web-scraper": {
      "command": "npx",
      "args": [
        "-y",
        "puppeteer-vision-mcp-serve"
      ],
      "env": {
        "OPENAI_API_KEY": "YOUR_OPENAI_API_KEY_HERE"
      }
    }
  }
}

Puppeteer Vision MCP Server 工具

工具列表暂未提供

当前收录暂时没有可展示的工具列表，不代表服务没有工具。请查阅项目文档了解支持的能力。

点击上方按钮,在临时沙箱中运行此服务器并实时获取工具列表。

概览

What is Puppeteer vision MCP Server?

This Model Context Protocol (MCP) server provides a tool for scraping webpages and converting them to markdown format using Puppeteer, Readability, and Turndown. It features AI-driven interaction capabilities to automatically handle cookies, captchas, and other interactive elements. It is intended for developers and LLM orchestrators that need structured web content.

How to use Puppeteer vision MCP Server?

Run the server directly via npx -y puppeteer-vision-mcp-server after setting required environment variables (e.g., OPENAI_API_KEY). It can also be configured as an MCP tool in an orchestrator using the JSON snippet provided in the README. Transport can be stdio (default), SSE, or HTTP, selected via the TRANSPORT_TYPE variable.

Key features of Puppeteer vision MCP Server

Scrapes webpages using Puppeteer with stealth mode
AI‑powered handling of cookies, CAPTCHAs, and overlays
Extracts main content with Mozilla's Readability
Converts HTML to well‑formatted Markdown with Turndown
Supports code blocks and tables with custom rules
Option to view browser interaction by disabling headless mode

Use cases of Puppeteer vision MCP Server

Scrape news articles that require dismissing cookie consent banners
Extract documentation or blog posts from pages with paywalls or login prompts
Capture structured content (code blocks, tables) from dynamic web pages
Feed cleaned Markdown into an LLM for further processing or summarization

常见问题

What AI model does it use?

It requires an OpenAI API key; the default vision model is `gpt-4.1`. You can change the model via the `VISION_MODEL` environment variable, and point to alternative providers by setting `API_BASE_URL`.

How do I see the browser during scraping?

Set the environment variable `DISABLE_HEADLESS=true` to run the browser in visible (non‑headless) mode.

What transport modes are supported?

Three modes are supported: `stdio` (default), `sse`, and `http`. Set the `TRANSPORT_TYPE` environment variable to switch between them.

Can I use a different AI provider?

Yes, set the `API_BASE_URL` environment variable to any OpenAI‑compatible

浏览器自动化分类下的更多 MCP 服务器

browser-use-mcp-server

co-browser

Browse the web, directly from Cursor etc.

827

Nullprint

nullprintai

Let Claude Code, Claude Desktop, OpenAI Codex, Kimi and any MCP-capable AI agent create, launch, check and drive browser profiles in Nullprint, the AI antidetect browser. Install with one command.

Fastcrw

Fast, lightweight Firecrawl/Tavily alternative in Rust. Web scraper, crawler & search API with MCP server for AI agents. Drop-in Firecrawl-compatible API (/scrape, /crawl, /search). 2.3x faster than Tavily, 1.5x faster than Firecrawl in 1K-URL benchmarks. 6 MB RAM, single binary.

286

Fetcher MCP

jae-jae

MCP server for fetch web page content using Playwright headless browser.

1.1K

HTML/CSS to Image

An MCP server for generating images from HTML & CSS or screenshots of URLs using htmlcsstoimage.com.