llms-txt-from-website

Yes, it automatically detects and reuses an existing /llms.txt manifest on the target website to minimize redundant work. If an existing llms.txt is found during the website documentation crawling process, the skill incorporates it directly rather than regenerating the content from scratch.

When direct repo discovery fails, the skill falls back to parsing sitemap.xml or crawling the website to generate a high-signal manifest. This ensures website documentation can still be converted into a structured llms.txt format even without explicit GitHub source links.

The llms.txt file is a curated manifest of documentation sources, while llms-full.txt is a structured concatenation of those sources with attached context. Generating llms-full.txt produces a comprehensive markdown file designed to provide maximum context for downstream LLM processing.

No, a sitemap.xml is not required because the skill first attempts to discover the documentation source repo via GitHub links. Sitemap.xml parsing is only used as a fallback mechanism when direct repository discovery is not possible during the website crawling process.