Quick answer: If you're on Shopify, you almost certainly already have an llms.txt. Since May 2026, every Shopify store serves /llms.txt, /llms-full.txt, and /agents.md at the root automatically — with /agents.md as the canonical file the other two mirror. So this isn't really a "how to add" job; it's a "how to check, customize, and not get fooled" one. And the honest part: no major AI answer engine has confirmed it uses llms.txt for rankings — Google has said outright it won't help — so treat it as low-effort future-proofing, not a traffic button.
That last point matters, because most guides still sell llms.txt as a magic "get into ChatGPT" file. The data disagrees. In a June 2026 study of 137,210 domains, Ahrefs found that 97% of published llms.txt files received zero requests in a month, and actual AI answer-bots (like PerplexityBot and OAI-SearchBot) made up only ~1.1% of the requests that did happen (Ahrefs). Their verdict was blunt: for showing up in ChatGPT or AI Overviews, an llms.txt file is "largely decoration."
So let's do this properly — check what you already have, customize it the supported way, put the right things in it — and then spend your real energy on the levers that actually move AI visibility.
Do you already have an llms.txt on Shopify? (Probably — here's how to check)
Open a browser and go to yourstore.com/llms.txt and yourstore.com/agents.md. On a live Shopify store in 2026, both return a real file — you didn't have to do a thing.
Here's what changed: in its May 28, 2026 changelog, Shopify made these files a native part of the platform (Shopify). Every store now serves three agent-discovery files at the root by default:
/agents.md— the canonical document (the one Shopify treats as the source of truth)./llms.txtand/llms-full.txt— alternate URLs that mirror/agents.mdunless you say otherwise.
They're served at your bare primary domain (no /en or Markets locale prefix), and there's no localized version — one file, one store. If you never touch a thing, Shopify keeps a sensible default in sync with your store settings. For a lot of merchants, "you already have one and it's fine" is the honest, complete answer.
What is llms.txt — and how is it different from agents.md and llms-full.txt?
llms.txt is a proposal from Jeremy Howard (co-founder of Answer.AI), published in September 2024 (llmstxt.org). The idea: give a language model a clean, Markdown map of your site's important content, so it doesn't burn its limited context window wading through nav, popups, and ad markup.
The spec has a specific shape:
- an H1 with your site/brand name (the only required part),
- a blockquote summarizing what the site is,
- optional detail, then
- H2 sections, each a list of
[link](url): descriptionentries — with an## Optionalsection for links a model can skip when it needs to be brief.
A vocabulary note so you're not confused later: in the wider spec, a "full" file means all the linked content inlined into one long document. On Shopify, /llms-full.txt is simply another URL that mirrors /agents.md by default — a narrower meaning. And agents.md is the emerging "agent instructions" convention that Shopify chose as its canonical file. In practice, on Shopify you customize agents.md and the other two follow.
How to customize your Shopify llms.txt / agents.md (the supported way)
If the default isn't detailed enough, the correct, Shopify-supported way to change it is a theme template — not an app, not a redirect. The exact path:
- Shopify admin → Online Store → Themes.
- On your live theme → ⋯ → Edit code.
- In the Templates folder → Add a new template.
- Create
agents.md.liquid(this controls all three files at once — the best default), orllms.txt.liquidonly if you want/llms.txtto differ from/agents.md.
Two rules that trip people up:
- These must be Liquid templates, not JSON — same as
robots.txt.liquid. You write Markdown with Liquid mixed in. - Shopify gives the template a special
agentsobject with auto-filled values — use them instead of hardcoding URLs so nothing goes stale:agents.store_name,agents.store_url,agents.sitemap_url, and (for agentic commerce)agents.ucp_discovery_urlandagents.mcp_endpoint_url(Shopify docs).
Shopify serves the first file it finds, in this order:
/agents.md→ youragents.md.liquid→ Shopify's default/llms.txt→ yourllms.txt.liquid→ youragents.md.liquid→ Shopify's default
So adding just agents.md.liquid upgrades all three in one move.
What should you put in it?
Think of it as a concise, curated map for a shopping agent — not a data dump. A good Shopify agents.md / llms.txt includes:
- Your store name (H1) and a one-line blockquote: what you sell and who for.
- A pointer to your sitemap (
agents.sitemap_url). - Key collections — your main category pages (the "shop this catalog" links).
- A short list of best-seller / hero products (curated, not your whole catalog).
- Policy links — shipping, returns/refunds, warranty.
- FAQ, support, contact, and About pages.
- If you're doing agentic commerce: your UCP discovery and MCP endpoint via the
agentsobject. - Push non-essential links under an
## Optionalheading (per the spec).
Honest caveat: a rich, hand-curated file is genuinely useful for coding/IDE agents and is nice future-proofing — but there's no evidence today that it increases how often ChatGPT, Gemini, or Perplexity cite your store. Keep it tidy; don't spend a week on it.
Does llms.txt actually work in 2026? (The honest answer)
Short version: not as a ranking or citation signal for the big AI answer engines. The record:
- Google is explicit. Gary Illyes said in 2025 that "Google doesn't support llms.txt and isn't planning to," and Google's own documentation (updated June 2026) states you don't need machine-readable AI files to appear in Google Search or its AI features, because Google doesn't use them — llms.txt "won't harm or help" your rankings (Search Engine Land).
- The traffic data agrees. The Ahrefs study above found 97% of
llms.txtfiles get zero requests, and AI answer-bots barely fetch them. - Who does read it: coding and IDE agents (Claude Code, Cursor, Copilot). Tellingly, OpenAI, Anthropic, and Perplexity all publish an
llms.txton their own developer-docs sites — useful for docs and coding agents, not proof their answer engines crawl yours. - Some SEO posts claim Perplexity and Anthropic "use" your
llms.txtfor retrieval. We couldn't verify that against a primary source, and the server-log data points the other way — so treat it as reported, not confirmed.
We go deeper on the "does it work" debate in how to get your Shopify store recommended by ChatGPT, Gemini & Perplexity. The takeaway here: keep your llms.txt because it's free and tidy — just don't expect it to move traffic.
The old workarounds (app, URL redirect, CMS page) — skip them
If you're following an older tutorial, ignore these — they're now redundant or actively broken on Shopify:
- A "Page" called llms lives at
/pages/llms-txt, which is not/llms.txt. Bots check the root; they'll never find it. - URL redirects or App Proxy files pointing
/llms.txtelsewhere now get overridden — Shopify's platform routing owns that path. (Somellms.txtapps built on App Proxy are effectively superseded.) - The only external override that still works is putting Cloudflare in front of your store to intercept the request first — an engineering project, not merchant guidance.
For 99% of stores: use the default, or customize the theme template. That's it.
So what actually gets your store recommended by AI?
If llms.txt is decoration, where should your effort go? The levers actually confirmed to help AI shopping assistants find and trust your store:
- Clean, complete product data — titles, descriptions, specs, and image alt text an AI can parse.
- Schema.org (JSON-LD) — valid Product, FAQ, and Breadcrumb markup so machines know what each page is.
- AI-crawler access — make sure GPTBot, ClaudeBot, PerplexityBot, and Google-Extended aren't blocked in your robots settings.
- Reviews and trust signals — the social proof AI leans on when it decides what to recommend.
- Shopify's Agentic Storefronts / UCP — Shopify's own agentic-commerce plumbing (part of the Universal Commerce Protocol) that surfaces eligible products inside AI shopping (Shopify Engineering).
Auditing all of that by hand — across a whole catalog — is real work. That's the gap our app fills: AIRank Xpert scores every product for AI-readiness, bulk-fixes titles, descriptions, tags and alt-text plus Schema data, checks that AI crawlers can actually reach you, and — the part nothing else does — tracks your AI Share of Voice across ChatGPT, Gemini, Perplexity and Claude week over week, so you can see whether any of this is working. Note what it doesn't claim: it won't "add llms.txt" (Shopify already does that) or promise a spot in ChatGPT — it strengthens the levers that matter and measures the outcome, honestly. See how AIRank Xpert works → (free plan available).
Shopify's three agent files at a glance
| File | What it is | How to customize | Fallback |
|---|---|---|---|
/agents.md | Canonical agent-discovery doc | agents.md.liquid | Shopify default |
/llms.txt | Concise LLM index (mirrors agents.md) | llms.txt.liquid → agents.md.liquid | Shopify default |
/llms-full.txt | Alternate URL (mirrors agents.md) | llms-full.txt.liquid → agents.md.liquid | Shopify default |
Sources
- llms.txt — the specification (llmstxt.org)
- Shopify — Customize /llms.txt, /llms-full.txt and /agents.md (changelog, May 2026)
- Shopify — agents.md.liquid template docs
- Ahrefs — 97% of llms.txt files are never read (study, June 2026)
- Search Engine Land — Google says llms.txt won't harm or help rankings
- Shopify Engineering — Universal Commerce Protocol (UCP)
Related reads:
- Shopify AEO: how to show up in AI answer engines
- How to get your Shopify store recommended by ChatGPT, Gemini & Perplexity
- How to make your Shopify store buyable in ChatGPT (agentic commerce)
- Agentic storefronts + Shopify Catalog, explained
Xpertshire builds AI-discovery-ready Shopify stores, structured-data fixes, and AEO/GEO content. We also publish AIRank Xpert — which scores your products for AI-readiness, fixes the data and schema that actually matter, and tracks whether ChatGPT, Gemini and Perplexity are recommending you.



