article-extractor

Use the reader CLI or trafilatura to parse the URL and output only the main content. Both tools strip navigation, ads, and signup forms, producing clean text saved to a file named after the article title.

reader uses Mozilla's Readability algorithm and works well for most articles, while trafilatura is more accurate for news sites, blogs with complex layouts, and non-English content. The skill tries reader first, then falls back to trafilatura.

Yes, a fallback method uses curl plus a Python HTMLParser script to extract paragraph and heading content. It is less reliable than reader or trafilatura and may include some noise, but works without dependencies.

Extraction fails when articles sit behind paywalls or logins, or when sites render content with heavy JavaScript that curl cannot execute. In those cases the tools return empty content and the skill reports the failure.

The article title is cleaned by removing special characters like slashes, colons, and quotes, then truncated to a safe length. The result is saved as a .txt file, with a preview of the first lines shown for verification.