Skip to content
Free audit · No sign-up to start

Check whether AI understands your store

We check prices, availability, shipping and returns on your public pages, and show you what to fix first.

We analyse the domain you enter and show you the result as soon as it finishes.

What gets checked

Six categories, each with its weight in the score.

The engine evaluates your domain's public pages and splits the result into six weighted categories. What cannot be observed does not score: it is left unevaluated and does not count as a failure, not filled in with a guess.

25 %

Catalogue

Readable product pages: clear title, image and data an agent can quote.

weight: 25 %
20 %

Trust

Returns, shipping, contact and privacy published where they can be found.

weight: 20 %
15 %

Navigation

Title, description and categories that let you reach a product without guessing.

weight: 15 %
15 %

Commercial

Price, variants and a detectable add-to-cart step.

weight: 15 %
10 %

Technical

Accessible robots.txt, canonicals and valid structured data.

weight: 10 %
15 %

Agent protocols

llms.txt, discovery manifests, headers and agentic commerce signals.

weight: 15 %
How it is measured

What this audit measures and what it does not

19 checks in 6 categories, on what an agent can see from outside your store. What cannot be observed does not score: you are told which checks those are and they do not count as a failure.

What it measures

  • Whether agents can get in: robots.txt, sitemap and files meant for them (llms.txt and similar).
  • Whether they can understand what you sell: structured data, prices, availability and policies visible on public pages.
  • Whether the store announces how to work with agents (public protocols and manifests).

What it does not measure

  • It makes no test purchases and adds nothing to the cart: it does not check that checkout works.
  • It does not test payments, agent identity or your store's internal rules.
  • It does not enter private areas: no accounts, passwords or session cookies.
  • It does not run JavaScript: what only appears once a browser executes it is not seen.

Crawling: we read robots.txt first and, if it forbids crawling, we do not crawl. We request at most 15 public pages and identify ourselves as AgenticMCPStores-AuditBot/1.0 (+https://trusteed.xyz/audit); you can block us in your robots.txt.

This score is a one-off analysis of your website. It is not Trusteed's signed Trust Score nor what you see in your platform's panel: they are different products with different data, and they are not mixed.

After the audit

The audit is a snapshot. The Trust Score is the continuous signal.

The audit measures what an agent finds on your site today. The Trust Score measures the operation, is signed and can be queried.

Agent-ready audit: what an agent sees when it reaches your store