Site files
Generate llms.txt, sitemaps, JSON-LD and Markdown for your own site, and check that the site serves what you generated.
Technical SEO → Site files builds files for your own site from the pages of your last site audit. You put each file on your site; the product never writes to it. It then fetches the published path and tells you whether the live file matches.
What it makes
| File | How it is built |
|---|---|
llms.txt | Name and description from your confirmed brand facts (the home page description until you confirm one), then the audit's indexable pages: home first, then by path depth, at most 100. It helps agents read your site. No effect on AI citations has been proven, so none is claimed. |
sitemap.xml | The audit's pages that returned 200, are not noindex and do not point their canonical elsewhere. lastmod appears only where the product observed a page change. For a target engine's limits: Google and Bing 50,000 URLs per file, Baidu 50,000 URLs and 10 MB, Toutiao 10,000 URLs and 10 MB. A larger site gets an index file plus numbered parts; each part also stays under 5 MB, the most the product stores and can read back to check. |
| JSON-LD | Product, Offer, Organization, Article, Service, MerchantReturnPolicy and OfferShippingDetails from the fields you give. Nothing is produced while a required field is missing. FAQPage is not offered: Google stopped showing FAQ rich results on 2026-05-07. |
| Markdown | One page of your site as clean Markdown. Navigation, forms and scripts are dropped; links keep only http and https. Nothing is stored. |
A file built from an audit that stopped early, or that filled its page budget, is marked partial with the number of pages it saw. Add the rest from your own sitemap.
Checking the live file
Check on my site fetches the path from your own host (the www twin counts as the same
site), obeys your robots.txt for AIEOBot, and refuses a redirect to another site. The result is
one of matches, differs (with the first line that differs), missing (404 or 410) or could not
check. Line endings and trailing spaces do not count as a difference. These fetches use the same
daily request budget as the audit and are limited to 10 an hour per user.
Each project keeps at most 30 generated files and 50 MB. Generating the same path again replaces the file and clears its check.
Structured data in the audit
The audit now reads each page's JSON-LD and reports, for the types it knows (Organization, LocalBusiness, Product, Offer, Article, BreadcrumbList, Service, WebSite):
- missing required fields (warning) and missing recommended fields (notice);
- JSON-LD that does not parse;
- a price in the structured data that the page does not show next to a currency sign or code. Google requires structured data to match what visitors can see.
Pages from an audit made before this check are shown as not checked, never as passed. The audit also notes buttons, links and form fields an agent cannot name, and clickable elements that are not marked as buttons. Google's guidance is to use semantic HTML; we do not claim an effect on visibility.