Agent Product Graph

Our crawler

Crawling is disabled by default and requires per-merchant opt-in. Most data comes from feeds merchants supply directly.

Identification

Where crawling is enabled, requests identify themselves as AgentProductGraphBot/0.1 with a link to this page and a contact address.

What we honour

  • robots.txt, per RFC 9309, including Allow/Disallow precedence, wildcards and Crawl-delay.
  • Your sitemap, including the Sitemap: directive in robots.txt. We prefer being told what you have over guessing from category pages, and we use lastmod so an unchanged page is not re-crawled.
  • Retry-After on 429 and 503 responses.
  • Conditional requests using ETag and Last-Modified, so an unchanged page is not re-downloaded.
  • One concurrent request per domain by default, with a configurable delay.

What we never do

  • Bypass a CAPTCHA, a login, a paywall or any bot protection.
  • Use stealth or anti-detection tooling, or rotate proxies to evade a block.
  • Ignore robots.txt, or treat a 5xx response from it as permission.
  • Follow a sitemap entry pointing at another site. A sitemap speaks for its own domain and nothing else.
  • Continue after repeated 403 or 429 responses — the source pauses itself automatically.

Opting out

Contact us from an address on the domain through the developer support form, or add a Disallow rule for our user agent. We will also honour a request to remove data already collected.