Submit your tool

Checklist

Is your store ready for AI shopping agents? 18 checks for find, read, and checkout

Eighteen checks for a store an agent has to find, read, and pay. Mark a row only when you can point at the feed, the HTML, or the contract. The score is a worksheet. It is not a crawl, and it is not an order.

OpenAI’s crawler page and Perplexity’s crawler page were read . Visa’s production-readiness page and Mastercard’s October 20, 2026 Agent Pay for Machines post were read the same day. The feed fields and the UCP flag are the ones in the store-readable guide, read November 10, 2026. Protocol install counts are the November 11 census. This page did not re-count stars.

Score: 0 of 100. Answered weight: 0. Still blank: 100 points across 18 rows.

100 points are still blank. The number above counts only the rows you marked.

Yes scores the full weight. Partly scores half. No scores zero. The weights are this desk’s. Nothing you mark is stored, and the answers are not sent with the page view.

Can an agent find a product row?

A homepage sentence cannot. The agent needs a feed row, or HTML it can quote, and a crawler you did not block.

Can an agent read the price and the stock?

The price and the stock have to be in the HTML response, on a stable URL, in the same shape the feed uses.

Can an agent check out?

Guest checkout is a store choice. UCP and ACP are channel integrations. Visa, Mastercard, and AP2 are not a theme setting.

Which gaps cost the most points?

No gap yet on the rows you marked. Blank rows are not counted as failures.

What does the score count?

It counts the rows you marked, using weights that sum to 100. A feed, a Merchant Center approval, and a price in the HTML are worth more than a crawler token or an llms.txt file.

Yes takes the full weight. Partly takes half. No takes zero. Blank rows stay out of the gap list so an unfinished worksheet does not look like eighteen failures. The weights are editorial. Google and OpenAI do not publish this score.

Can an agent find a product without a feed?

ChatGPT’s merchant path reads a product feed. A crawl of the homepage is optional.

OpenAI’s feed spec, read on this site November 10, 2026, names nine discovery fields: item id, title, description, product URL, brand, seller name, image URL, availability, and price. Search eligibility defaults to on. Checkout defaults to off. The merchant page says shopping is live for ChatGPT users in the United States, Shopify and Etsy catalogs are already integrated, and the purchase finishes on the merchant’s website or app. The longer field rules are how to make your store readable by AI shopping agents.

Which crawler puts a store in ChatGPT search?

OAI-SearchBot. Sites that opt out are not shown in ChatGPT search answers.

OpenAI’s crawler overview, read November 12, 2026, says OAI-SearchBot surfaces websites in ChatGPT search, a robots.txt change can take about 24 hours to apply, and GPTBot is a separate training crawler. ChatGPT-User fetches when a person asks. Robots.txt rules may not apply to it, and it does not decide Search. PerplexityBot, on Perplexity’s crawler page read the same day, surfaces and links sites in Perplexity search. Perplexity-User is the user-triggered visit. Googlebot remains the crawler for Google Search. Google-Extended is the control for training and grounding in some of Google’s other systems, not the switch for AI Overviews, on Google’s AI features page as read on this site November 9.

Does llms.txt or JSON-LD turn on checkout?

Neither one authorizes a payment. Both can still make a price easier to quote.

Google’s AI features page says you do not need a new machine-readable file or extra schema.org markup for AI Overviews or AI Mode, and that structured data should match the visible words. llmstxt.org is a separate proposal for a Markdown index at /llms.txt. This site publishes its own copy. A Product node that repeats the HTML price gives an agent a second copy of a number it can check against the feed. It does not set Google’s native_commerce flag.

Which checkout protocol can a store turn on today?

The one the channel you sell on will certify. Stars do not pick it.

Google’s UCP guide says checkout eligibility defaults to false until the feed sets native_commerce to true on an eligible SKU. OpenAI’s is_eligible_checkout does nothing until checkout is enabled for that integration. The November 11 census found no official npm or PyPI package for ACP or for Google’s Agent Payments Protocol. AP2’s README still pointed at a git install, and the PyPI name ap2 belonged to a different repository. UCP’s latest GitHub release in that census was v2026-08-25. ACP’s newest dated spec folder was spec/2026-04-17.

Visa Intelligent Commerce, on the production-readiness page read November 12, 2026, is sandbox, then certification, then production for an approved participant. Mastercard Agent Pay, announced April 29, 2025, is tokenized payment for registered agents. Agent Pay for Machines, opened to early access October 20, 2026, is a different product: an agent paying for digital services on x402, inside a budget. Who can pay on the shopper products already shipping is can an AI shopping agent check out for you. The dated company calendar is AI shopping agents, the next 6 months. The shopper table is Best AI agent for shopping.

What should you do with the gap list?

Fix the largest missing weight first, then re-mark the row. Start with the feed and the HTML price.

A missing OpenAI row or a price that exists only after JavaScript costs more points than a missing llms.txt, because those two are what a harness quotes. After the row and the price agree, set the checkout flag only on SKUs the channel’s guide allows. Then write the eval: did the charged amount match the approved amount, did the item id match the feed, and did the run stop when the only path was a handoff. That eval is a harness question. The worksheet for the software around the model is the harness evaluation checklist. Live counts of a different package set are the trackers. To list an agent product, submit a tool or claim a page in the directory. More worksheets are on the tools index.

Which pages were read for these 18 checks?

The crawler pages, Visa’s production path, and Mastercard’s October 20 post were read November 12, 2026. Feed fields and the UCP flag are the November 10 reading. Install counts are the November 11 census.

Source: aiagentsinfo.com, Is your store ready for AI shopping agents? 18 checks for find, read, and checkout, updated .

Cite or embed this

The citation names this page and the date above. The embed is a blockquote with a link back here. Nothing here is a required badge.

Citation

AI Agents Info. "Is your store ready for AI shopping agents? 18 checks for find, read, and checkout." Sources read November 12, 2026. https://www.aiagentsinfo.com/tools/shopping-agent-readiness-checker.

Embed

<blockquote cite="https://www.aiagentsinfo.com/tools/shopping-agent-readiness-checker">
<p>Score whether an AI shopping agent can find a product, read the price and stock, and check out. 18 weighted checks, a gap list, and a fix for each. Nothing is stored. Sources read November 12, 2026.</p>
<footer><a href="https://www.aiagentsinfo.com/tools/shopping-agent-readiness-checker">Is your store ready for AI shopping agents? 18 checks for find, read, and checkout</a>, AI Agents Info.</footer>
</blockquote>

What else do merchants ask before an agent can quote the store?

The score is a worksheet of eighteen sourced checks. It does not save answers and it does not place an order.

What is an AI shopping agent readiness checker?

Eighteen yes, partly, or no checks a merchant answers about its own store. The score is a weighted worksheet for whether an agent can find a product row, read the price and stock in the HTML, and check out through a channel that has actually been turned on. It does not crawl the store.

Does a high score mean an agent can buy from the store?

No. Ninety points with every row answered means the worksheet is full. OpenAI’s merchant page, read on this site November 10, 2026, still sends the ChatGPT purchase to the merchant’s site or app. Google’s checkout flag defaults to false. Visa Intelligent Commerce and Mastercard Agent Pay are participant programs, not a theme setting.

Are the weights an official score?

No. The 18 weights sum to 100 and were set by this desk. A feed row and a price in the HTML are weighted above llms.txt and Google-Extended, because those two files are not how ChatGPT or Google Search decide to show a product. Partly scores half the row.

Is the worksheet saved?

No. The answers stay in the browser until you leave or clear them. There is no account, no email field, and no file download. Copy puts the gap list on your clipboard.

Which checkout protocol should a store turn on first?

The one the channel you already sell on will certify. UCP is the Google Merchant Center flag, and it stays false until you set it on an eligible SKU. ACP’s feed flag does nothing until OpenAI checkout is enabled for that merchant. AP2, as of the November 11 census, is a git install with no official registry package. Visa and Mastercard require their own onboarding.