Amazon scraper API for product, offer and search data

An Amazon scraper API is a hosted service that fetches public Amazon pages for you and hands back the content, so your code never talks to Amazon directly. The hard parts of scraping Amazon are not the HTTP request. They are keeping requests from being blocked, rendering pages from the right country, and keeping a parser working when the page layout changes. A scraper API moves all three to the provider.

Roketfy's version goes one step further than returning HTML. Four of the five Amazon endpoints return structured JSON for a specific page type: product detail, seller offers, search results and reviews. The fifth returns raw HTML for any Amazon URL, for the cases where you want to do your own parsing.

You need this kind of API when the data you want belongs to listings you do not own. Amazon's own APIs are tied to a seller or affiliate account and its terms. A scraper API reads the public storefront, so it works for any ASIN on the marketplace you name.

Call `/v1/amazon/product-detail` with an `asin`, the 10-character product id found in every Amazon product URL. The response is the full product page as structured data: title, pricing, images, specifications and description. This is the endpoint to use as an Amazon product API when you need one record per product.

The optional `geocode` parameter sets the marketplace by country code, such as us, gb, de or jp, so the same ASIN can be read as a shopper in each country sees it. A postal code or country-level location narrows it further for marketplaces where price and availability depend on the delivery address, and `language` sets the page language.

If you want to check the parsed fields against the source, `include_html` returns the page HTML alongside the JSON in the same call.

The product page shows one price. `/v1/amazon/offers` shows all of them: every seller offering the ASIN, with price, shipping, condition, fulfilment type and which offer holds the Buy Box. It costs 1 credit, the same as product detail, and takes the same `asin` and location parameters.

This is the endpoint most price monitoring setups are built on. Keep a list of the ASINs you care about, call offers for each on your own schedule, and store the result with a timestamp. Comparing two snapshots tells you who changed price, who entered or left the listing, and when the Buy Box moved.

Yes. `/v1/amazon/search` takes a `keyword` and returns the search results with the listings and their ranking. Category pages use the same structure on Amazon, so a category term works here too. A `page` parameter moves through result pages, starting at 1.

Search is the usual starting point when you do not have ASINs yet. Run a keyword, collect the ASINs from the results, then pass them to product detail or offers. Run on a schedule, the same call gives you a rank history for a keyword in a given marketplace.

`/v1/amazon/raw-html` takes a full Amazon URL and returns the complete rendered HTML of that page, from the location you name. The response is `{ html }` rather than structured JSON, and the page is capped at 4 MB upstream.

Use it for pages the structured endpoints do not cover, or when you already have a parser and only need reliable fetching. You keep control of extraction and take on the maintenance that comes with it. For product, offer and search pages, the structured endpoints save you that work at the same 1 credit per call.

Reviews work differently from the other four endpoints. `/v1/amazon/product-reviews` costs 10 credits per page and requires your own signed-in Amazon cookies, passed as `cookies`. Amazon's review pager answers a signed-in session and nothing else, so a call without cookies is refused rather than answered with an empty page. Each page returns rating, title, text, date, author and helpful votes, and the endpoint supports batch tasks for collecting many pages. The Amazon reviews API page covers the cookie setup and paging in detail.

Product detail, offers, search and raw HTML cost 1 credit per successful call and need no Amazon account. Each of them accepts an optional `super` flag that routes the call through residential and mobile proxies for a higher success rate, at 10 credits instead of 1.

Credits are bought as pay-as-you-go packages, and failed requests are not charged. All five endpoints are also available as MCP tools, so an AI agent can call them with the same token and credit balance as your backend. Full parameter lists are in the docs at data.roketfy.com/docs.