How it works

When someone types a store address, Catalog Truth makes one short, read-only visit to that store only: its public product list (/products.json), up to 20 product pages, its policy pages, its contact page (and, only if that is missing, /pages/contact-us and its home page) and robots.txt. That is at most 29 requests, three at a time. It never follows links to other sites, never submits forms, and never crawls stores on its own.

It identifies itself as CatalogTruth/0.1 (+https://catalogtruth.tryneverclosed.com/about; one store check requested by a visitor). A store can be checked at most once every 30 minutes; repeat requests get the saved result.

Every finding comes from fixed rules comparing two things the store itself publishes. No AI model decides anything.

It follows robots.txt the way Google documents it, using the rules for CatalogTruth if a store has them and the rules for all crawlers otherwise. Pages a store asks crawlers to skip are never fetched; the report lists them as not checked. A store that adds User-agent: CatalogTruth with Disallow: / is never checked at all.