- Methodology
- Parameters
- llms.txt Presence
llms.txt Presence
experimentalCategory: ai-signals · Methodology v4.5
/llms.
Signal Source
- Source
https://{domain}/llms.txt (fallback /llms.md, /.well-known/llms.txt)- Kind
- http_response
Score Bands
| Verdict | Condition |
|---|---|
| Pass | a valid llms.txt is fetched and follows the expected markdown structure — a # title, a > description blockquote, and at least one ## section |
| Partial | an llms.txt (or fallback) file is fetched but its markdown is incomplete — missing the title, the description, or any ## section |
| Fail | no llms.txt is found at /llms.txt, /llms.md, or /.well-known/llms.txt, or the response is empty |
Description
What this parameter measures
/llms.txt is an emerging standard that gives AI systems a markdown summary of what your site offers without crawling every page. This parameter checks whether you publish one. friendly4AI fetches /llms.txt first, then falls back to /llms.md and /.well-known/llms.txt. When a file is found, it validates the basic structure: a top-level # title, a > description blockquote, and at least one ## section heading. It also counts the sections and the markdown links inside the file as supporting evidence.
Why it matters for AI-readiness
A well-formed llms.txt is a curated front door for AI systems. Instead of inferring your purpose from scattered pages, an LLM reads a single concise document that names your site, summarizes what it does, and links to the pages that matter most. That shortcut improves how accurately ChatGPT, Claude, and Perplexity describe and cite you, and it gives you direct control over the framing rather than leaving it to a crawler's heuristics. As an emerging convention, llms.txt is still gaining adoption. This parameter is the methodology's experimental AI-signal, so it is published with that status while the standard stabilizes.
How we score it
This experimental AI-Specific Signals parameter is a gradient scored across three tiers under the v4.4 methodology, and the scanner enforces the published rubric exactly. It passes (100) when a file is fetched and is valid — meaning it has a # title, a > description, and at least one ## section. It scores partial (50) when a file is fetched but the markdown is incomplete, missing any of those three required elements. It fails (0) when no file is found at /llms.txt, /llms.md, or /.well-known/llms.txt, or the response is blank. Section and link counts are recorded as evidence but the pass gate is the three-element structure check.
How to fix common issues
- Create
/llms.txtat your site root in markdown: a# Your Site Nametitle on the first line, a> One-line summaryblockquote, and one or more## Sectionheadings. - Link to your most important pages from within the sections so AI systems can follow them to deeper content.
- If your file scores partial, check that all three required elements are present — a missing description blockquote or section heading is the most common cause.
- Use
/llms.txtas the primary path; the/llms.mdand/.well-known/llms.txtfallbacks exist for compatibility but the root file is the convention. - See the llmstxt.org specification for the full format, then re-scan to confirm the file validates.
Version History
- Introduced
- v1
- Last changed
- v4.4
Key takeaways
- Signal: https://{domain}/llms.txt (fallback /llms.md, /.well-known/llms.txt)
- Category: AI-Specific Signals
- Passes when: a valid llms.txt is fetched and follows the expected markdown structure — a #…
- Status: Experimental — scoring behaviour may change in future methodology versions.