What problem does it solve? Turning arbitrary web pages, PDFs, YouTube videos, and Word documents into clean article text with accurate citation metadata is error-prone: extraction returns empty bodies, authors and dates come out wrong, and PDF dependencies leak into bundles. This Skill guides you through the extract-webpage package so you can pick the right call, diagnose failures, and fix citation or extraction issues. ## Core Features & Use Cases - URL-to-cited-article pipeline: extractContent() accepts a URL, HTML/Markdown string, DOM Document, or DOCX buffer and returns the article body plus title, author, date, source, and a formatted APA citation. - Fetching and extraction control: scrapeURL handles bot detection, proxies, and redirects; Readability and Mercury extractors can be run independently; site-specific rules come from fetchScrapingRules. - Keyphrase and summary extraction: extractSEEKTOPIC produces keyphrases and top sentences via an LLM path or an n-gram/TextRank fallback, with query-biased summarization. - Use Case: A research tool needs the body text and a properly formatted citation for any link a user pastes, including PDFs and YouTube videos, without bundling pdfjs into every consumer's build. ## Quick Start Ask the assistant to extract the article content and citation from a given URL using extractContent with a raised timeout, and inspect the returned error field if the result is empty.
