Scrapling crawling framework: 88,000 star open source crawler, automatic adaptation of webpage revision also comes with MCP for AIScrapling 抓取框架：8.8 万星开源爬虫，网页改版自动适配还自带 MCP 给 AI 用

Traditional crawlers are most afraid of website revisions: the front-end is adjusted, the selector is completely invalid, the script is scrapped on the spot, and manual troubleshooting and rewriting are required.

The core innovation of Scrapling is the adaptive selector: it learns the structural characteristics of the target element, automatically repositioning the page according to the similarity after the page changes, and the redesign is no longer equal to rewriting.

With this mechanism and out-of-the-box anti-climbing capability, the project has accumulated 88,000 stars on GitHub and has dominated the trend list for a long time.

Lightweight mode: single request plus intelligent content extraction, suitable for grabbing interfaces, taking structured data, extremely fast.

Dynamic mode: Built-in headless browser that automatically handles JavaScript rendering, scroll loading, and captcha scenarios.

Scale: Concurrent scheduling, automatic retry, request concealment and fingerprint disguise out of the box, single request and whole station crawling use the same set of writing, from demo to production without changing the framework.

The fastest way: pip install scrapling A command is installed (the official PyPI source code package included with this article can be used for network inconvenience, only 0.18 MB, pip install local files are installed offline).

For AI: It has a built-in MCP server. After configuring the scrapling mcp command in MCP clients such as Claude Desktop, the AI can grab web pages and parse data by itself.

The first time you run the recommended scrapling install to complete the browser component, the dynamic mode will be fully available.

The download button provides the PyPI official source scrapling version 0.5 source code package (tar.gz format, only 0.18 MB). Execute pip install scrapling-0.5.tar.gz to install offline.

The project is licensed in BSD style and commercially friendly; please comply with the robots agreement of the target website and local regulations to capture data, and use it only for legitimate purposes.

This article is compiled from public sources and ships with the matching resource. Product features and pricing are subject to the official page. Resources are for learning and exchange only — please respect the original license.