Getting a GitHub Pages site cited by AI
The five things a report asks for, and where each lives
Checked 2026-09-14. Platforms change what they expose; if one of these is out of date, it is out of date visibly rather than quietly.
- Page titles & meta descriptionsYes
Front matter or the template head, depending on your generator.
- JSON-LD schemaYes
An include in the layout.
- robots.txtYes
A file at the repository root, published with the site.
- A plain-text file like /llms.txtYes
A file in the repository, served as text/plain by extension. A project site publishes at /repo-name/, and llms.txt v2 defines exactly that case: a file at a subpath covers the pages under it — GitHub Pages is the example the spec cites.
- New pages (FAQ, comparison, use case)Yes
The normal page editor.
No server-side redirects, which matters for consolidation
GitHub Pages serves static files; there is no 301 rule you can write. If the report finds the same content reachable at several URLs, the usual fix — redirect the duplicates to one canonical URL — is not available. What is available: a canonical link tag on each page, and a redirect plugin for your generator that emits meta-refresh stubs. Both are weaker than a 301, so it is worth not creating the duplicates in the first place.
Why there is no writer for this platform
No writer is planned. The changes are commits, and you already have a review process for those — which is the same thing the approval queue being built does for everyone else.
Either way the report splits its actions into two columns: what we publish for you and what only you can do — claiming a directory profile, fixing a Google Business listing, answering the questions a page has been avoiding. The second column has checkboxes, because a re-test run before you have done those is a re-test measuring nothing.
Where to start on GitHub Pages
None of this needs an account with us. The AI crawler checker and the Website metadata checker answer the technical questions in about a minute.
- 01
Add llms.txt and robots.txt to the repo root
Two files in one commit.
- 02
Put schema in the layout, not in each page
One include, rendered everywhere.
- 03
Set canonical tags deliberately
They are the strongest consolidation signal you have here.
Questions
See what the engines say about your GitHub Pages site
12 buyer questions across ChatGPT, Perplexity and Gemini. The diagnosis is the same whatever the site is built with — only the fixing differs, and we say which half is which.