crawl-mcp

crawl-mcp

MCP server for web crawling, searching, and AI-powered content extraction, supporting single-page, batch, and full-site crawling along with text, news, image, book, and video search.

Category
Visit Server

README

crawl_mcp

基于 crawl4ai 和 FastMCP 的 MCP 服务器,提供网页爬取和 AI 分析功能。

PyPI Version GitHub

功能

爬取工具

  • crawl_single - 爬取单个网页,返回 Markdown 格式(浏览器渲染,适合 SPA)
  • extract_url - 轻量级 URL 提取(无需浏览器,速度快 5-10 倍)
  • crawl_site - 递归爬取整个网站
  • crawl_batch - 批量爬取多个网页(异步并行)

搜索工具

  • search_text - 通用网页搜索
  • search_news - 新闻内容搜索
  • search_images - 图片搜索(支持下载和 AI 分析)
  • search_books - 图书/电子书搜索
  • search_videos - 视频搜索(含时长、播放量等)

AI 能力

  • LLM 集成 - AI 驱动的内容提取和摘要(先快速爬取,后可选处理)
  • 自动重试 - 网络错误自动重试(指数退避)

LLM 处理设计

爬取和 LLM 处理分离,确保快速响应:

  1. 快速爬取(6-10秒)- 始终返回原始 Markdown
  2. 可选后处理 - 如提供 llm_config,对 Markdown 进行 AI 处理

性能对比

场景 耗时 说明
extract_url(静态页面) ~1.5s ddgs extract,无需浏览器
crawl_single(无 LLM) ~7s 浏览器渲染
crawl_single(有 LLM) ~40s 爬取 + AI 处理
crawl_batch 2 页(无 LLM) ~15s 并行爬取
search_text / news / books / videos ~1.5-2s ddgs 搜索

关键优势

  • 静态页面用 extract_url,1.5 秒出结果
  • SPA/JS 重度页面用 crawl_single,浏览器渲染保证完整
  • 搜索类工具全部基于 ddgs,秒级响应

安装

pip install crawl-mcp

MCP 配置

Claude Desktop

{
  "mcpServers": {
    "crawl-mcp": {
      "command": "uvx",
      "args": ["crawl-mcp"],
      "env": {
        "OPENAI_API_KEY": "your-api-key"
      }
    }
  }
}

高级配置(可选)

如需自定义 API 端点或模型:

{
  "mcpServers": {
    "crawl-mcp": {
      "command": "uvx",
      "args": ["crawl-mcp"],
      "env": {
        "OPENAI_API_KEY": "your-api-key",
        "OPENAI_BASE_URL": "https://api.openai.com/v1",
        "LLM_MODEL": "glm-4.7",
        "VISION_MODEL": "glm-4.6v"
      }
    }
  }
}

环境变量

变量 说明 默认值
OPENAI_API_KEY API 密钥 必填
OPENAI_BASE_URL API 基础 URL https://api.openai.com/v1
LLM_MODEL 文本模型名称 glm-4.7
VISION_MODEL 图片分析模型名称 glm-4.6v

LLM 配置

所有工具支持可选的 llm_config 参数:

{
  "instruction": "提取产品信息",
  "schema": {
    "type": "object",
    "properties": {
      "name": {"type": "string"},
      "price": {"type": "number"}
    }
  }
}
  • instruction: 提取指令
  • schema: 可选的 JSON Schema

注意: api_keybase_urlmodel 从环境变量读取。

搜索功能

extract_url - 轻量级 URL 提取

适用于静态页面、文章、博客等不需要 JS 渲染的场景。

{
  "name": "extract_url",
  "arguments": {
    "url": "https://example.com/article",
    "fmt": "text_markdown"
  }
}

参数说明

  • url: 要提取的网页 URL
  • fmt: 输出格式(可选)
    • text_markdown: Markdown 格式(默认)
    • text_plain: 纯文本
    • text_rich: 富文本
    • text: 原始 HTML
    • content: 原始字节

search_text - 通用网页搜索

适用于搜索技术文档、百科、博客、论坛、教程等静态内容。

{
  "name": "search_text",
  "arguments": {
    "query": "Python 快速排序算法",
    "region": "cn-zh",
    "max_results": 5
  }
}

参数说明

  • query: 搜索关键词
  • region: 区域代码(可选)
    • wt-wt: 无区域限制(默认)
    • us-en: 美国(英语)
    • cn-zh: 中国(中文)
    • uk-en: 英国(英语)
    • jp-jp: 日本(日语)
  • safesearch: 安全搜索(可选)
    • on: 严格过滤
    • moderate: 适度过滤(默认)
    • off: 关闭过滤
  • timelimit: 时间限制(可选)
    • d: 最近一天
    • w: 最近一周
    • m: 最近一月
    • y: 最近一年
  • max_results: 最大结果数(默认:10)

返回格式

{
  "success": true,
  "query": "Python 快速排序算法",
  "count": 5,
  "results": [
    {"title": "...", "href": "https://...", "body": "..."}
  ]
}

search_news - 新闻搜索

适用于搜索突发新闻、时事、财经、体育等时效性内容。

{
  "name": "search_news",
  "arguments": {
    "query": "人工智能最新进展",
    "timelimit": "w",
    "max_results": 10
  }
}

参数说明

  • search_text 相同,但 timelimit 仅支持 dwm(不支持 y

返回格式

{
  "success": true,
  "query": "人工智能最新进展",
  "count": 3,
  "results": [
    {
      "date": "2024-07-03T16:25:22+00:00",
      "title": "...",
      "body": "...",
      "url": "https://...",
      "image": "https://...",
      "source": "..."
    }
  ]
}

search_images - 图片搜索

搜索图片,支持下载到本地和 AI 分析。

{
  "name": "search_images",
  "arguments": {
    "query": "cute cat",
    "max_results": 10,
    "download": true,
    "download_count": 5,
    "analyze": true,
    "analysis_prompt": "描述这张图片的内容和风格"
  }
}

参数说明

参数 说明 默认值
query 搜索关键词 必填
region 区域代码 wt-wt
max_results 搜索结果数量 10
size 图片尺寸 -
color 颜色过滤 -
type_image 图片类型 -
layout 布局方式 -
download 是否下载到本地 false
download_count 下载数量 全部
output_dir 下载目录 ./downloads/images
analyze 是否 AI 分析 false
analysis_prompt 分析提示词 详细描述这张图片的内容

图片过滤选项

  • size: Small, Medium, Large, Wallpaper
  • color: Red, Orange, Yellow, Green, Blue, Purple, Pink, Black, White, Gray, Brown, Monochrome
  • type_image: photo, clipart, gif, transparent, line
  • layout: Square, Tall, Wide

返回格式

{
  "success": true,
  "query": "cute cat",
  "search_results": {
    "count": 10,
    "results": [
      {
        "title": "...",
        "image": "https://...",
        "thumbnail": "https://...",
        "url": "https://...",
        "width": 1920,
        "height": 1080,
        "source": "Bing"
      }
    ]
  },
  "download_results": {
    "total": 5,
    "downloaded": 5,
    "failed": 0,
    "output_dir": "./downloads/images"
  },
  "analysis_results": {
    "count": 5,
    "results": [
      {
        "image": "...",
        "type": "local",
        "analysis": "这是一张可爱的猫咪图片..."
      }
    ]
  }
}

search_books - 图书搜索

适用于学术研究、技术书籍整理、文献检索。

{
  "name": "search_books",
  "arguments": {
    "query": "clean code software engineering",
    "max_results": 5
  }
}

返回字段:title, author, publisher, url, thumbnail, info

search_videos - 视频搜索

适用于教程搜集、视频素材整理、内容分析。

{
  "name": "search_videos",
  "arguments": {
    "query": "python tutorial beginner",
    "max_results": 5
  }
}

返回字段:title, duration, embed_url, thumbnail, statistics (viewCount), publisher, uploader

开发

uv sync
uv run pytest
uv run python -m crawl4ai_mcp.fastmcp_server --http

许可证

MIT License

Recommended Servers

playwright-mcp

playwright-mcp

A Model Context Protocol server that enables LLMs to interact with web pages through structured accessibility snapshots without requiring vision models or screenshots.

Official
Featured
TypeScript
Magic Component Platform (MCP)

Magic Component Platform (MCP)

An AI-powered tool that generates modern UI components from natural language descriptions, integrating with popular IDEs to streamline UI development workflow.

Official
Featured
Local
TypeScript
Audiense Insights MCP Server

Audiense Insights MCP Server

Enables interaction with Audiense Insights accounts via the Model Context Protocol, facilitating the extraction and analysis of marketing insights and audience data including demographics, behavior, and influencer engagement.

Official
Featured
Local
TypeScript
VeyraX MCP

VeyraX MCP

Single MCP tool to connect all your favorite tools: Gmail, Calendar and 40 more.

Official
Featured
Local
graphlit-mcp-server

graphlit-mcp-server

The Model Context Protocol (MCP) Server enables integration between MCP clients and the Graphlit service. Ingest anything from Slack to Gmail to podcast feeds, in addition to web crawling, into a Graphlit project - and then retrieve relevant contents from the MCP client.

Official
Featured
TypeScript
Kagi MCP Server

Kagi MCP Server

An MCP server that integrates Kagi search capabilities with Claude AI, enabling Claude to perform real-time web searches when answering questions that require up-to-date information.

Official
Featured
Python
E2B

E2B

Using MCP to run code via e2b.

Official
Featured
Neon Database

Neon Database

MCP server for interacting with Neon Management API and databases

Official
Featured
Exa Search

Exa Search

A Model Context Protocol (MCP) server lets AI assistants like Claude use the Exa AI Search API for web searches. This setup allows AI models to get real-time web information in a safe and controlled way.

Official
Featured
Qdrant Server

Qdrant Server

This repository is an example of how to create a MCP server for Qdrant, a vector search engine.

Official
Featured