Artifact endpoints

Once the integration is mounted, these URLs are served from your domain. That is the whole point: content addressed at your site earns citations for your site.

On your domain	What it is	Served by
/llms.txt	The agent document: what an AI system should read first about your catalogue. One body, three names, because three conventions exist	your three root-file routes
/llms-full.txt
/agents.md
/apps/geoffy/llm/{page path}.md	One page, as plain text, on a site with no commerce platform. The page /de/programmes/data-science has its copy at /apps/geoffy/llm/de/programmes/data-science.md, so two pages that share a handle each have their own	the namespace route
/apps/geoffy/llm/{handle}.md	One product, as plain text, on a site with Shopify behind it. On a site with no commerce platform this address also answers, with the page whose path is /{handle}, else the page with that handle that was published last
/apps/geoffy/guides/{slug}	One published buying guide
/apps/geoffy/guides/{slug}.md	The same guide, as plain text
/apps/geoffy/sitemap.xml	Every Geoffy surface on your domain, listed
/robots.txt	Your own rules, with the Geoffy crawler block merged in	your robots route
/{key}.txt	Your IndexNow key, so index pings for your product pages are accepted	a static file you add

Your product pages themselves carry the widget and the structured data, rendered by the component you mounted.

What you will see in your own logs

Condition	Your route answers
Everything is working	200, cached for your revalidateSeconds
Nothing is published for that page, or a Shopify store's own site is not verified yet	404. Geoffy serves published content only. There is no draft view
An old address of a buying guide that moved	301 with a Location under /apps/geoffy/ on your domain, never one on Geoffy's domain
Geoffy is unreachable or degraded	503 with a Retry-After, never a fabricated 404. A crawler retries a 503; a 404 tells it to drop the page from its index, which is a far more expensive answer to give on your behalf
Your product page, when Geoffy has nothing to add	Renders exactly as it would without us. The helpers resolve to null rather than throwing, and every caller renders nothing

Reading Geoffy directly while debugging

Your own caches sit between you and us, so when your site and your Geoffy dashboard disagree, check the source. Your dashboard shows the direct URL for each of your published surfaces. Fetching one bypasses every cache of yours at once.

If Geoffy and your mounted route disagree, the difference is in your cache, not in your integration.

Reading them while debugging

Your site’s settings in Geoffy show the direct URL for each of your published surfaces. Fetching one bypasses every cache of yours at once, which is the fastest way to tell whose side a problem is on.

This bypasses your Next.js cache entirely, including .next/dev/cache. If Geoffy and your mounted route disagree, the cache is at fault; see testing locally.