Happy slice of mcp pizzaMCP.pizza

crawl4ai-mcp-server

MCP.Pizza Chef: weidwonder

Two tools do the work: a free web search that needs no sign-up, and a page reader that strips out menus, headers and footers and returns clean text with the links kept intact. Google results are available as well, but only if you supply your own Google search key, which is free for a hundred queries a day and charged after that. One caveat: the 'trimmed' output option always comes back empty, because the filter it relies on is never switched on. Written in Chinese, last updated January 2026.

Web/Research

Use This MCP server To

Search the web without signing up for anything Turn a cluttered article into clean readable text Read a page my assistant cannot open by itself Gather a handful of sources on a topic at once Keep the links from an article for citing later

README

MseeP.ai Security Assessment Badge

Crawl4AI MCP Server

smithery badge

这是一个基于MCP (Model Context Protocol)的智能信息获取服务器,为AI助手系统提供强大的搜索能力和面向LLM优化的网页内容理解功能。通过多引擎搜索和智能内容提取,帮助AI系统高效获取和理解互联网信息,将网页内容转换为最适合LLM处理的格式。

特性

  • 🔍 强大的多引擎搜索能力,支持DuckDuckGo和Google
  • 📚 面向LLM优化的网页内容提取,智能过滤非核心内容
  • 🎯 专注信息价值,自动识别和保留关键内容
  • 📝 多种输出格式,支持引用溯源
  • 🚀 基于FastMCP的高性能异步设计

安装

方式1: 大部分的安装场景

  1. 确保您的系统满足以下要求:

    • Python >= 3.9
    • 建议使用专门的虚拟环境

  2. 克隆仓库:

git clone https://github.com/yourusername/crawl4ai-mcp-server.git
cd crawl4ai-mcp-server
  1. 创建并激活虚拟环境:
python -m venv crawl4ai_env
source crawl4ai_env/bin/activate  # Linux/Mac
# 或
.\crawl4ai_env\Scripts\activate  # Windows
  1. 安装依赖:
pip install -r requirements.txt
  1. 安装playwright浏览器:
playwright install

crawl4ai-mcp-server FAQ

Is it still being maintained?
Lightly. The last change was January 2026, and the project is openly a one-person weekend build. It works, but do not expect quick fixes.
Do I need a key or an account?
Not for the default search. DuckDuckGo results need nothing at all. Only if you want Google results do you have to create your own Google search key and paste it into a config file.
Does the Google option cost money?
Beyond a hundred searches a day, yes. Google bills for the extra queries against the key you supply, so the charge lands on you, not the project.
Can I use this to research a topic and read the sources?
Yes. Ask it to search, then ask it to read the promising results; each page comes back as tidy text with the original links preserved.
Why does the condensed output come back blank?
Because the trimming feature is never actually enabled in the code, so that option returns an empty result. Stick to the default format, which already strips navigation, headers and footers.
Sometimes it finds nothing at all. Is it broken?
Possibly not. When the search source rate-limits it, the code quietly returns an empty list rather than an error, so a blocked search looks identical to no results.
How hard is the setup?
Genuinely technical. You clone the code, make a Python environment, install dependencies and download a full browser engine, which is a few hundred megabytes. A Smithery one-liner exists for Claude Desktop.
Is the documentation in English?
No. The readme is written in Chinese, so you may need to translate it while following the install steps.