llms.txt
An optional emerging navigation file that some AI-oriented tools may choose to interpret.
llms.txt is best treated as a concise, human-readable inventory published at /llms.txt. A store can use it to point willing software toward canonical explanations, policies, and guides. That is a modest documentation benefit—not evidence of higher AI visibility, guaranteed ingestion, or adoption by any named engine.
Its value depends on honest scope. Keep important information in crawlable HTML, manage permissions through robots.txt, and maintain normal search signals. An independent agent may choose to request llms.txt, ignore it, or interpret it differently. The publisher offers the map; the recipient decides whether and how to use it.
What is llms.txt?
llms.txt is an optional text file through which a publisher presents a curated map of important site resources. The llms.txt proposal site suggests Markdown headings, short descriptions, and links so a willing agent can find useful context. The file itself does not confer authority, access, indexing, or ranking.
Think of it as an editorial contents page for software. It can highlight documentation that already exists, but should never become the only location of a fact shoppers need.
Is llms.txt an official web standard?
No. llms.txt is a public proposal, not an Internet Standard, search-engine directive, or universally negotiated protocol. By contrast, the Robots Exclusion Protocol is documented in RFC 9309. Calling both files “instructions for bots” hides a decisive difference in status and function.
The proposal may evolve, and independent agents may experiment with it. Unless a vendor’s current official documentation says otherwise, do not claim that vendor reads, honors, or ranks llms.txt.
What does Google Search say about llms.txt?
Google’s AI optimization guide, updated July 10, 2026, explicitly says Google Search ignores llms.txt: the file neither helps nor harms visibility or rankings. From Google Search’s perspective, AEO and GEO optimization remain SEO, with no special file required for its AI experiences.
Google also says no special schema, chunking, or AI-only writing is needed and warns against separate pages for every query fan-out. Its SEO Starter Guide and AI features guidance remain the relevant Google sources. These statements describe Google Search, not every independent agent.
How does llms.txt differ from robots.txt, sitemaps, canonicals, and structured data?
Each artifact has a separate job: robots.txt governs crawler permissions; sitemap.xml supports URL discovery; canonical and meta robots signals guide indexing treatment; structured data describes visible entities; llms.txt offers an unofficial inventory. Substituting one for another creates technical errors and usually leaves the intended control unmet.
| Artifact | Primary job | What it does not do |
|---|---|---|
| robots.txt | Expresses crawler permissions under REP | Does not guarantee deindexing or ranking |
| sitemap.xml | Lists preferred URLs for discovery | Does not force crawling or indexing |
| Canonical and meta robots | Signal duplicate consolidation and index treatment | Do not grant crawler access |
| Structured data | Describes entities represented on the page | Does not guarantee a search feature |
| llms.txt | Curates links for willing agents | Does not control access, indexing, or ranking |
What should an llms.txt file contain?
A useful llms.txt should contain only a short site identity, a plain description, and links to durable canonical resources with specific labels. Favor policies, product guidance, methodology, definitions, and support documentation. Avoid keyword blocks, copied page bodies, fabricated claims, temporary campaign URLs, or instructions that contradict the linked pages.
# Example Store
> Product documentation and customer policies.
## Key pages
- [About](/pages/about): Company and sourcing facts.
- [Shipping](/policies/shipping-policy): Delivery terms.
- [Returns](/policies/refund-policy): Return conditions.
## Product guidance
- [Size guide](/pages/size-guide): Measurement method.Serve the example at the root /llms.txt. Plain text is sensible, but the proposal is not a negotiated HTTP standard; normal HTTP semantics still govern the response.
How do you add llms.txt to Shopify?
Publish the file only if the exact root path /llms.txt can return the intended text reliably. Shopify themes do not automatically turn an arbitrary theme asset or page handle into that root response. Use a supported app, proxy, edge, or hosting arrangement, then test the public result instead of assuming deployment succeeded.
The StoreCited llms.txt generator prepares a concise draft. Follow the Shopify installation walkthrough for routing choices, while keeping every referenced fact available on normal HTML pages.
How should you validate and maintain llms.txt?
Validation should prove only that the file is reachable, readable, current, and linked to valid canonical resources. It cannot prove ingestion or a visibility effect. Record the response and date, check every destination, and repeat those checks after migrations, policy changes, redirects, or major content releases.
- Select a small set of durable, canonical resources.
- Draft accurate labels and descriptions without promotional claims.
- Publish the text at the exact root path.
- Confirm a successful response and readable body.
- Test every listed destination for redirects or errors.
- Check canonicals, meta robots rules, and crawl permissions.
- Save a dated copy and named content owner.
- Revalidate after platform, URL, or policy changes.
Assign the file to a named content owner and store its review date beside the release record. For each change, note the reason, affected links, approver, and deployment identifier; then retain the previous version so regressions can be traced. A quarterly review is a practical default for stable documentation, while policy or URL changes should trigger immediate validation. This creates auditable maintenance evidence without implying engine adoption.
The StoreCited AI crawler checker can inspect public access signals. That observation is distinct from whether an external agent later fetches or uses llms.txt.
When is llms.txt worth using?
llms.txt is worth considering when a site already has strong canonical HTML, stable documentation, and someone responsible for keeping a small inventory current. It is low priority when essential pages are blocked, duplicated, inaccurate, thin, or difficult to navigate. Fix those foundational problems before polishing an optional map.
Maintenance cost is the real test: stale links weaken the file’s documentary value. Use the structured data glossary when entity markup is the actual need, or run a free StoreCited scan to observe crawl, content, and schema readiness signals available at scan time. The scan does not monitor engines, guarantee citations, or establish causality.