← all toolsTool
llms.txt Intelligence Analyzer
Don't just generate /llms.txt; grade the one your site has. Fetches /llms.txt and /sitemap.xml from your domain, parses both, and scores six axes deterministically: title, tagline, description, section coverage (Docs, Pricing, Contact, About), sitemap consistency, and link counts. Returns a suggested replacement body with every gap closed.
Tool
Examples
Click an example to run the tool with pre-filled inputs.
Why this matters for AI agents
A weak /llms.txt is worse than none. If agents fetch a file with no description and stale links, they cite the stale ones and skip the real pages. The bar to clear is low (title, tagline, four sections, real URLs), but most published files miss two or three of those.
This tool runs the same checks AgentSpeed runs against llms_txt.present, standalone, against any URL.
Common fixes
Title missingAdd `# {Brand}` as the first line. Agents cite this as the canonical identifier; without it, they invent one from your homepage title tag.
3+ expected sections missingDocs, Pricing, Contact, and About are the four sections agents expect. If any are missing, the suggested replacement at the bottom of this tool fills them in.
Links not in sitemapWhen /llms.txt and /sitemap.xml disagree on which URLs are canonical, agents and search crawlers fragment. Either add the /llms.txt links to the sitemap, or pin /llms.txt to URLs already in it.
Tagline missingA `> one-liner` blockquote helps agents quote your positioning verbatim. Skip it and you get paraphrases.
Many off-domain linksA few external citations are fine; majority-off-domain looks like a content farm. Keep the link list focused on your own canonical pages.
Linked rubric checks
llms_txt.presentWhether /llms.txt is present at the apex with at least an H1 and a markdown link.
llms_txt.presentsitemap.presentWhether /sitemap.xml exists; the analyzer cross-references it for consistency.
sitemap.present- markup.canonical
FAQ
Don't I just need the llms.txt generator?
The generator builds a new file. The analyzer grades the file you have. Most teams ship an /llms.txt once and forget about it; this tool is for catching drift between what you wrote and what agents now expect.
Does the score check that links are reachable?
Not in v1. Verifying every link would 30× the request count and bog down the tool. We compare /llms.txt links against /sitemap.xml as a proxy: if the link is in the sitemap, it is presumed reachable. Dead-link verification ships in v2 once we have a budget for it.
How is the score computed?
Deterministic, no model. Each axis subtracts a fixed amount from 100 when violated: missing title -15, missing tagline -5, missing description -5, missing each expected section -8, no links -20, fewer than 3 links -8, sitemap mismatch -8. Bounded 0–100.
Why these four sections?
Docs, Pricing, Contact, About cover the four canonical questions agents are asked: how do I use it, what does it cost, how do I reach support, who builds it. Other sections (API, Changelog, Status, Policies) are useful but not required for a passing grade.
Run a full agent-readiness scan
Full scanScore the entire site against 30 agent-readiness checks across discoverability, readability, structured data, actionability, and performance.
Get the full report →