Hi everyone,
Important heads up: if your store has a custom llms.txt set up: Shopify just claimed the /llms.txt platform-wide and is redirecting it to a new agents.md file at the root of your site.
That means if you were using a proxy, redirect, file or app to serve your own llms.txt, it’s being overwritten right now.
Visit [yourstore.url]/llms.txt right now to check what’s actually being served!
The good news: you can still take control of what these files say, and add in relevant AI context and store content as structured signals for improved AI visibility.
I made a short video here breaking down what changed and exactly how to fix it
What do you think of the change? Drop any questions below.
Hey @Eligijus !
The file size limit for any file whether LLMS.txt or Agents.md should be 256 Bytes, no? We also have a workaround to make sure all the data can be added there if it goes over.
I’m also confused with what you mean by workaround as this redirect is baked in directly at the server level. You can still create those template files, but the agent will simply be redirected the moment they hit the /llms.txt.
It’s 256KB, but when you deal with stores over 5k products - that’s not enough.
And while you cannot get the urls at /llms.txt and /llms-full.txt - you can do It another way.
I’ll not share what we found here for everyone to copy us, but there is a solution haha
So this solution shows the LLMS.txt or it doesn’t? I’m confused 
Like send me a Shopify Store URL where the llms.txt and llms-full.txt shows up to show me that your solution works
Worth stepping back from the redirect mechanics for a second. The harder question is whether AI crawlers actually consume /llms.txt enough to make this fight worth it.
Server logs from a few stores I help with show GPTBot, PerplexityBot, ClaudeBot all hitting /robots.txt, /sitemap.xml, and crawling product pages directly. The /llms.txt hits are real but tiny, and I haven’t seen evidence that content there shows up in citations vs what they pull from the rendered HTML + JSON-LD.
If Shopify is forcing agents.md and capping it, the practical play is probably:
- accept the agents.md as the bare-bones brand context
- invest the energy in clean JSON-LD (Product, FAQPage, Article, BreadcrumbList) since that’s what crawlers reliably parse
- make sure sitemap.xml is fresh and product pages render server-side
Curious if anyone has correlation data showing /llms.txt content actually surfacing in AI Overviews or chat citations? I haven’t been able to make that link.
@lumine I completely agree that llms.txt can seem overhyped at times, and LinkGPT for example only treats llms.txt as one component in a broader visibility framework that includes JSON-LD and LLM Outreach. LLMS.txt alone is likely not enough to drive the needle significantly.
However, there are some major signals:
Google itself has included llms.txt to Lighthouse agentic browsing scoring recently: Lighthouse agentic browsing scoring | Chrome for Developers
Maintained llms.txt files also now ship from Vercel, Anthropic, Stripe, Cloudflare, Cursor
And while there is no public documentation confirming or denying because model ingestion pipelines are not transparent. We do know:
• LLM systems rely heavily on structured data
• Crawlers and retrieval systems prefer explicit machine-readable endpoints
• Every major search evolution has rewarded clarity and structure
There’s no clear direction because no defined methodology exists. Previously, we relied on tools like webmaster tools and sitemaps, but this approach lacks a centralized system making it both beneficial and challenging.
Hi @ryankatsnel,
I can see this is a massive move by Shopify to standardize AI signals, but it will be frustrating for those who already had custom setups. It’s clear they are prioritizing agentic commerce above all else.
Does Shopify provide a native way to customize the content of that new agents.md file yet, or is it currently a static template for everyone?
Thank you for sharing,
Worth separating two things here. The redirect is baked in at the server level, so /llms.txt and /llms-full.txt both bounce to /agents.md. That means agents.md is the only file we actually control.
But being able to overwrite agents.md doesn’t mean we can just paste the old llms.txt content at the bottom. They do different jobs. agents.md is the structured file that tells commerce agents how to act under UCP. llms.txt (and especially llms-full.txt) was a catalog/context map for crawlers. Dumping the catalog underneath Shopify’s agent instructions either buries those instructions or blows past the file’s size limit, since a full llms-full.txt is way too big for it.
So it’s not “you can’t,” it’s “don’t do it flat.” Keep Shopify’s directives intact, add a short brand summary up top, and link out to a separate catalog instead of inlining everything.
And +1 on JSON-LD: clean Product/FAQPage/BreadcrumbList markup and a fresh sitemap is where the crawlers actually look anyway.
Who said you can’t just paste llms.txt and llms-full.txt at the bottom ?
If LLMs start reading agents.md - they will see links to more information which they can use.
Shopify will once again correctly redirect to llms.txt and llms-full.txt, provided these files have been created at templates/llms.txt.liquid and templates/llms-full.txt.liquid. If these files do not exist, the content of agents.md will be displayed.
@Nordalux I just saw that and got an email from one of my users. Great news!
Thanks for the heads up. Just checked my store. Shopify is definitely serving their own llms.txt now. My custom version is gone.
The agents.md file is new. Looks like a replacement. You can edit it through the Shopify admin under settings. Not as flexible as a custom file but at least you have some control.
Worth adding that llms.txt and agents.md are crawl signals, not guarantees. They inform AI agents what’s available but they don’t control how ChatGPT, Gemini or Perplexity actually represent your brand when a customer asks a buying question. Those recommendations are shaped by a much wider set of signals; structured data, review aggregations, third-party mentions, FAQ Schema and how your product descriptions are semantically structured.
Fixing the file is definitely the right play, but merchants shouldn’t mistake it for full AI Visibility control. Most stores have no idea whether they’re appearing in AI recommendations at all, let alone whether what’s being said is accurate. But the structures content angle is definitely where the ranking signal lives.