Extract product prices, availability, images, identifiers and variants from public product pages. Discover URLs from store sitemaps, filter by data completeness, and export JSON or CSV. Free-plan sample: up to 25 pages per run.
Security: URLs (typed, sitemap-listed or redirected to) that resolve to localhost, private or cloud-metadata addresses, or use a non-http(s) scheme, are refused and reported as unreachable rows.
crawlerIdentity is now also used for product-page requests (it was only used for discovery).
Duplicate and malformed start URLs are dropped up front (duplicates were charged twice); numeric inputs given as strings are accepted.
A store whose discovery fails no longer aborts the whole run.
0.1 — unreleased
First public build.
One normalised row per product from JSON-LD, microdata and OpenGraph, with a
completeness score and named missingFields.
Product discovery from robots.txt and the sitemaps it declares, trusting the
store's own filing over path shape so flat-URL stores work.
ProductGroup variants kept whole: shared fields from the group, price and
stock from each variant.
Prices parsed to numbers in both of the world's separator conventions;
unreadable prices come back absent rather than as 0.
Non-UTF-8 pages decoded in the encoding the server declared.
Unreachable pages are recorded as ok: false rows and are never charged.
A run that reaches its charge limit stops and says so, rather than
continuing to fetch pages it can neither bill nor deliver.
