The Tavily Extract Node extracts the main content from one or more specified URLs. It strips away cookie banners, navigations, ads, and footers, returning raw, clean text or markdown.
Parameters
Required Parameters
- urls (array of strings): List of target URLs to extract content from.
Optional Parameters
- format (string): Output format. Choose
markdown(Default) ortext. - extract_depth (string): Use
advancedfor extraction from highly protected/complex sites (like LinkedIn) or pages with embedded table content. - query (string): A query string to rerank and extract specific chunks of text by relevance.
- include_images (boolean): Extract image asset URLs from the page.
- include_favicon (boolean): Extract favicon URLs.
Practical Use Cases
- Clean Scraper: Pull pure article text or documentation pages directly into your LLM's context without wasting tokens on HTML boilerplate.
- Reranked Retrieval: Provide a
queryto pull only the most relevant sections of a long website rather than the entire page. - Cross-source Document Syncing: Extract markdown content from diverse product pages to dynamically update internal knowledge bases.