Open-source, LLM friendly, crawler, scraper. Three of those four describe the tool. One describes the consumer on the other end, and that is the only one that changes what the output has to look like.
Calling output LLM friendly commits the tool to a shape, because the reader on the other end has a context window and no patience for navigation chrome.
The description pairs traversal and extraction in a single name, which sets the expectation that you do not assemble the two halves yourself.
No output format, no schema, no token counts, no site coverage, no sample page. The trending description names none of it.
Concept mock, not affiliated with the project. The GitHub Trending description on 30 August 2026 reads "Crawl4AI: Open-source LLM Friendly Web Crawler & Scraper" followed by a community invite link. The grading panel splits those four claim words and says which side of the tool each one describes. No output sample, benchmark, or format spec is represented here, because none is published in that line.