How it works
Paste your llms.txt content into the box and pick a crawler persona. The preview parses the file the same way that crawler would - extracting the H1, blockquote, primary source links, matched User-agent block and any citation hints - and shows you exactly which bits survive the trip into the model's context window. If a section is missing, the preview says so in plain language.
Why crawlers disagree
- GPTBot is strict. It reads the H1 and blockquote as the canonical site summary and ignores poorly-structured sections. If your H1 is missing, ChatGPT may not know what your site is.
- ClaudeBot cares about structure. Heading hierarchy matters - Claude uses headings as section anchors and honors Preferred citations blocks when writing attributed answers.
- PerplexityBot is link-first. Every markdown link in the file becomes a citation candidate, so the Primary sources block is where you want your best URLs.
Pair with
Use the LLMs.txt Tester to confirm the file is reachable, validate syntax with the LLMs.txt Validator, and generate a fresh template from the Template Builder. Strategy reading: how AI assistants choose citations.