Scrapling

About

Scraping and browser automation your agent can call

Version 0.5 (October 17) rebuilt the MCP server around complete browser workflows: fill forms, click, scroll, type and wait in one ordered sequence, read or search the current page without reloading, run JavaScript, take element or full-page screenshots, inspect recorded network requests and save and restore login state. It also adds CAPTCHA solving (hCaptcha, reCAPTCHA, Cloudflare Turnstile and others), leaner tool descriptions to save tokens and clearer errors for agents. Outside MCP you get stealthy fetchers, async sessions, a spider framework, `Response.markdown()` for RAG and a site-to-Markdown crawler. Pages are cleaned of scripts and hidden prompt-injection text before they reach the model. v0.5 has many breaking changes, so read the migration guide first, and respect each site's terms when you scrape. Made by Karim Shoair (@D4Vinci1).

Features

• MCP server with full browser workflows (forms, clicks, scrolls, waits)
• Clean Markdown output with hidden prompt-injection text stripped
• Save and reuse logins across runs
• Recorded network history you can search
• Adaptive parser that survives layout changes
• Spider framework and site-to-Markdown crawler for RAG

Install