What problem does it solve? Extracting content from modern websites is difficult: many sites use JavaScript rendering, lazy-loaded images, and Cloudflare protection, while Claude's 5MB image limit blocks large image analysis. This Skill crawls any URL locally and produces a clean markdown file with OCR text inserted at each image position. ## Core Features & Use Cases - 3-Tier Fetching: Automatically escalates from plain HTTP to stealth browser (Camoufox) to Playwright with scrolling, handling static sites, SPAs, and Cloudflare-protected pages. - Structured Data Extraction: Pulls JSON-LD, Next.js RSC payloads, and meta tags from SPA sites (Korean e-commerce like Olive Young, Coupang, Musinsa) alongside trafilatura-cleaned content. - Gemini OCR Inline: Analyzes up to 15 images per page (20MB each, bypassing Claude's 5MB limit) and inserts OCR results directly at image positions in the output markdown. - Use Case: Provide a competitor's product page URL and receive a markdown file saved to your workspace containing the full page text, structured product data (price, brand, category), and OCR-extracted text from all product images. ## Quick Start Ask Claude to analyze a URL, for example: analyze this competitor site https://example.com and save it to my resources folder.
