Scrapling is an open-source Python web scraping framework by Karim Shoair (GitHub: D4Vinci). It combines HTTP fetching, browser automation, parsing and crawling in one library. It gained about 7,500 stars in a week on GitHub's trending list and now has about 98,700 stars and 10,300 forks. It has no notable Hacker News discussion in the last 30 days, but reviews by scraping companies and on DEV Community keep putting it on their 2026 lists.
What it does well: its main idea is adaptive element tracking. You save an element's fingerprint once with auto_save=True; when the site later renames classes or moves things, a selector run with adaptive=True uses similarity scoring to find the closest match instead of breaking. Fingerprints are kept in a local SQLite file. The parser has a BeautifulSoup-like API with CSS, XPath, text and regex search. In benchmarks from the project's repository cited by ScrapingBee, parsing about 5,000 elements took 1.99 ms, against 2.01 ms for Parsel/Scrapy and 1,541 ms for BeautifulSoup with lxml; ScrapingBee calls those numbers synthetic and directional.
Three fetchers, one API: Fetcher makes fast HTTP requests through curl_cffi with browser-fingerprint impersonation; StealthyFetcher is a Playwright-based browser with stealth patches; DynamicFetcher runs a full Playwright browser for JavaScript-heavy pages. There is also a Scrapy-like spider with per-domain throttling, pause and resume, robots.txt compliance and JSON, CSV or XML export, plus a CLI, an interactive shell, an MCP server and an Agent Skill so AI agents can browse and extract data, and a Markdown output mode for RAG.
How to try it: it needs Python 3.10 or later. pip install scrapling installs only the parser; for the fetchers run pip install "scrapling[fetchers]" and then scrapling install to download the browsers. The code is at github.com/D4Vinci/Scrapling under the BSD-3-Clause licence.
Limits: the author calls it "a pretty young project" that is "a bit rough around the edges", according to ScrapingBee, which also notes small bugs, docs that drift between versions, fuzzy matching that can pick the wrong element, and browser fetchers that cost more CPU and RAM. Stealth fetching, including its Cloudflare Turnstile handling, is not a guaranteed bypass. The README says the library is for educational and research purposes and tells users to follow scraping and privacy laws, site terms and robots.txt. For website owners, it is also a reminder of how capable today's free scrapers are.
