Catalog readiness for AI shopping agents
AI shopping agents skip catalogs without crawler rules and complete product JSON-LD
Job: make the catalog machine-readable before an AI shopping agent tries to discover it: publish llms.txt, set GPTBot / ClaudeBot / PerplexityBot in robots.txt on purpose, and put complete Product JSON-LD (and Open Graph) on live product pages.
A readiness score is a checklist. It does not enroll the store in any agent storefront, and it is not a ranking or inclusion promise.
Scan those three surfaces, fix the gaps the score cites, then re-scan when theme or feed changes drop structured data.
Install, run readiness, fix the first flagged gap
The paid path on this page is a readiness pass on the live catalog, then one concrete fix - not a redesign of the temporary Fable LP.
- Install Agentic Catalog Readiness from the App Store.
- Run the readiness check on the catalog this guide covers.
- Open the first flagged gap (missing fields, weak titles, or incomplete attributes the app lists).
- Fix that gap in Admin, then re-run to confirm the flag clears.
Stay on thicken and readiness. Do not treat this block as a cue to rebuild Oct1 Fable landing pages.
Install Agentic Catalog Readiness → run check → fix first flagged gap
Demo
Short walkthrough of the app flow. Then follow the start path below.
llms.txt, robots.txt, and Product JSON-LD are separate gaps
| Check | What “ready” means here | Common miss |
|---|---|---|
llms.txt |
Public hint of which URLs an LLM-facing agent should prefer | Missing file |
robots.txt for GPTBot / ClaudeBot / PerplexityBot |
Allow or deny those crawlers deliberately | Accidental block or accidental open with no policy |
| Product JSON-LD + Open Graph | Complete Product structured data on the live PDP | Incomplete offers / identifiers; gallery-only pages |
robots.txt allow is not the same as llms.txt. Missing either is a readiness gap. Complete JSON-LD alone does not make the catalog “agentic.”
Do not confuse this with an access audit of AI apps that can read customer data (different guide). A crawler allow is not permission for an installed app to read orders.
Suggested path:
- Install Agentic Catalog Readiness.
- Scan
llms.txt, AI-crawler robots rules, and Product JSON-LD / OG on live PDPs. - Repair barcode/GTIN where the listing marks that path live. Treat preview-only fields as preview until unlocked on the live listing.
- Schedule scans if you need alerts when the score moves. Export per-product evidence on plans that include it.
Start path: score, then repair identifiers
- Open the Agentic Catalog Readiness listing (
apps.shopify.com/agentic-catalog-readiness) and install it. - Run a scan of
llms.txt,robots.txtrules for GPTBot, ClaudeBot, and PerplexityBot, and Product JSON-LD / Open Graph on live product pages. - Repair barcode/GTIN where the listing marks that path live. Treat price, availability, and ProductGroup as preview-only until a paid unlock if that is still how the listing describes them.
- Schedule daily or weekly scans if you need email alerts when the score changes. Export CSV evidence per product on the plans that include it.
The app scores public crawler and structured-data hygiene. It does not submit the catalog to any agent network.
What a readiness score checks vs what it does not
| Check | Catalog consequence |
|---|---|
llms.txt |
Public hint of which URLs an LLM-facing agent should prefer |
robots.txt for GPTBot / ClaudeBot / PerplexityBot |
Whether those crawlers are allowed or blocked on purpose |
| Product JSON-LD and Open Graph | Whether live product pages expose complete structured data |
| Policy, shipping, tax, domain | Store-level inputs to the same score, not a legal opinion |
| Agent storefront enrollment | Out of scope — not this page’s job |
Example public files to confirm (paraphrase, not a template you must copy):
/llms.txt
/robots.txt User-agent: GPTBot | ClaudeBot | PerplexityBot
product page JSON-LD @type Product
gtin / barcode, offers, availabilityWhat this page is not
- Not a ranking, traffic, or “listed in ChatGPT shopping” guarantee.
- Not an access audit of AI apps that can read customer data. That job is staff permission drift.
- Not EU green-claim copy. That path is generic environmental claims on product copy.
- Not a feed submission to Google Shopping or any marketplace.
What does “catalog ready for AI shopping agents” mean on this page?
It means a crawler can find the store’s machine-readable catalog rules and a product page exposes complete product structured data. The checklist on this page is llms.txt, robots.txt rules for GPTBot, ClaudeBot, and PerplexityBot, plus Product JSON-LD and Open Graph on live product pages. It is not a promise that any agent will recommend or buy the product.
Is a robots.txt allow enough without llms.txt?
No. robots.txt tells specific crawlers whether they may fetch. llms.txt is a separate, public hint about which URLs and context an LLM-facing agent should prefer. Missing either one is a readiness gap. Confirm what you actually publish at both paths.
Does complete JSON-LD make the catalog “agentic”?
No. JSON-LD completeness (Product, offers, identifiers) is one input to a readiness score. Policy, shipping, tax, and domain checks sit beside it. Barcode/GTIN repair may be live while price, availability, or ProductGroup stay preview-only until a paid unlock. Preview is not a live catalog fix.
Will a high readiness score get the store into ChatGPT shopping or similar?
No. A score is an internal checklist against crawler rules and on-page structured data. Agent storefronts, merchant of record, and feed partnerships are separate programs. This page is not a ranking, inclusion, or sales guarantee.
How often should I re-scan?
Whenever you change robots.txt, llms.txt, or product identifiers, and on a daily or weekly schedule if you need email alerts when the score moves. A one-time scan goes stale the next time a theme or feed drops JSON-LD.
Is this the same job as auditing AI apps that can read customer data?
No. Staff and installed-app scope risk is an access audit. Catalog readiness is public crawler and structured-data hygiene. Do not treat a robots.txt allow as permission for an installed app to read orders.
Where do I score crawler rules and product JSON-LD?
Use a readiness scan that checks llms.txt, AI-crawler robots.txt, and Product JSON-LD / Open Graph on live product pages, then exports per-product evidence. Agentic Catalog Readiness is the App Store listing this site already uses for that job. Open the listing, install, run a scan, then fix the gaps the score cites.
Related pages
- Agentic Catalog Readiness — live App Store listing
- Staff permission drift access audit — installed AI apps, not crawler rules
- EU generic environmental claims ban — product-copy wording
- Shopify apps hub
- Tools
Not an inclusion or ranking guarantee. Preview-only catalog fields are not live repairs. Confirm what you publish at /llms.txt and /robots.txt after you change them.