Why agents get markdown and Google gets HTML
Same substance, different format — the guarantee that makes agent serving useful for agents and safe with search engines.
Why agents want markdown
An AI agent reading your product page pays — in time, money, and accuracy — for every token it processes. A typical themed store page is tens of thousands of tokens of navigation, scripts, and markup wrapped around a few hundred tokens of facts. The gateway serves agents just the facts, front-loaded: title, price, availability, variants, shipping, returns. Smaller pages get read more reliably, quoted more accurately, and recommended more confidently.
The same-substance guarantee
The markdown is a format transformation of your canonical page — generated from the same source of truth your website renders from. Same prices, same stock status, same policy text. No requester ever gets different substance: not a better price for agents, not different availability for anyone.
Why this is not cloaking
“Cloaking” — showing search engines different content than users see — is what gets stores penalized. Rebilder is built to stay on the right side of that line in two ways. First, the substance never differs, only the format. Second, known search crawlers are exempt from format switching entirely: Googlebot is classified as a crawler and receives your canonical HTML even if it explicitly asks for markdown. The markdown path is unreachable for search crawlers by construction, and a user-agent string alone never changes what content a request receives — only, at most, its format.