nocode-amazon

A Python client for ScrapingBee's three dedicated Amazon endpoints, plus the helpers that separate paid placements from earned ones.

The package name comes from the no code Make and Airtable workflow this grew out of. The library is the code path for people who outgrew the spreadsheet: same data, same public Amazon pages, no visual scenario to maintain.

Verified against the live API on 2026-09-10. Every parameter name, response field and credit figure below came back from a real call. Where the documentation and the API disagreed, the API won and the difference is noted.

pip install nocode-amazon

Requires Python 3.8 or newer and requests.

Authentication

Header based, on every request:

Authorization: Bearer YOUR_API_KEY

The api_key query parameter still answers but the current documentation marks it deprecated. This client sends the header.

from nocode_amazon import AmazonScraper

bee = AmazonScraper("YOUR_API_KEY")

Get a key and 1,000 free credits from ScrapingBee. Full reference: Amazon API documentation.

---

Endpoint reference

search(query, pages=1, **params)

GET /api/v1/amazon/search. 5 credits per page.

Amazon search results as structured JSON. No HTML, no selectors.

page = bee.search("fitness tracker")
page["products_count"]   # 23
page["products"]         # list of product dicts
page["refinements"]      # Amazon's own facets for this query
page["url"]              # the Amazon URL that was fetched
page["page"]             # 1

Live call on fitness tracker returned 23 products and a refinements object with 53 facet groups, including brands, band_color, band_material_type, battery_average_life, battery_charge_time and case_diameter.

Optional parameters, all verified present on the endpoint:

Parameter	Notes
pages	Pages to fetch. Billed per page
sort_by	featured, most_recent, price_low_to_high, price_high_to_low, average_review, bestsellers
category_id	Restrict to one Amazon category
merchant_id	Restrict to one seller
domain	Marketplace top level domain: com, co.uk, de, in
country	Proxy geolocation
currency	ISO 4217 display currency
language	ISO language code
device	desktop or mobile
zip_code	Postal code. Changes prices and delivery promises
light_request	Default True at 5 credits. False forces a browser at 15
add_html	Include the raw page alongside the JSON
autoselect_variant	Pick a variant automatically
screenshot	Capture the page. Always 15 credits, ignores light_request
tag	Your own label, returned in the response headers

Related landing pages: Amazon search API, Amazon keyword scraper API, Amazon organic results API, Amazon filters API, Amazon related searches API, Amazon spell check API, Amazon zip code API.

product(asin, **params)

GET /api/v1/amazon/product. 5 credits.

The wire parameter is query, not asin. The value has to be a valid 10 character ASIN, but the field is named query. This client takes asin as its argument name and sends query for you.

detail = bee.product("B0GTMTZF3V")
detail["brand"]          # 'Fitbit'
detail["price"]          # 99.99
detail["currency"]       # 'USD'
detail["rating"]         # 4.3
detail["reviews_count"]  # 2036

The live response carried 56 top level keys. The ones worth knowing:

Field	Contents
bullet_points	The feature bullets, newline separated
category	A ladder of breadcrumb steps, each with name and url
featured_merchant	name, seller_id, shipped_from, is_amazon_fulfilled
product_details	Spec table as key and value pairs, including best_sellers_rank
technical_details	The second spec table
rating_stars_distribution	Review counts per star rating
sales_rank	Category rank entries
variations	Other sizes, colours and configurations
images	Image URLs
delivery	Delivery promises with type and date
pricing_url	The offer listing page for this ASIN
parent_asin	The variation parent
discount_percentage, price_strikethrough, coupon	Promotion state
stock, max_quantity	Availability

Related: Amazon ASIN API, Amazon image API, Amazon review API, Amazon best sellers API, Amazon video results API, Amazon URL API.

pricing(asin, **params)

GET /api/v1/amazon/pricing. 5 credits.

This endpoint takes asin on the wire. Search and product both take query. Pricing does not, and it rejects query explicitly:

{"errors": {"query": {"asin": ["Missing data for required field."], "query": ["Unknown field."]}}}

That rejected call was billed 0 credits, confirmed by spb-cost: 0 on the response. This client raises ScrapingBeeError with the API's own error dict in .payload, so the offending field is never a guess.

offers = bee.pricing("B0GTMTZF3V")
for offer in offers["pricing"]:
    print(offer["seller"], offer["price"], offer["condition"])

Each offer carries seller, seller_id, seller_link, condition, price, price_shipping, currency, rating_count and a delivery_options list.

Related: Amazon offers API, Amazon seller API, Amazon vendor API, Amazon shipping API.

usage()

GET /api/v1/usage. Free.

bee.usage()
# {'max_api_credit': 1000000, 'used_api_credit': 353946,
#  'max_concurrency': 100, 'current_concurrency': 0,
#  'renewal_subscription_date': '2026-09-21T12:06:28'}

Call this before a large run. Note it lags by minutes, so do not read it immediately afterwards to compute what a batch cost.

---

Placement helpers

Amazon search markup gives every card the same classes, so CSS selectors cannot tell a paid slot from an earned rank. The endpoint labels each result instead, which turns the problem into a filter.

Per product placement fields:

Field	Type	Meaning
is_sponsored	bool	Paid placement
sponsored_position	int or None	Rank among the ads
organic_position	int or None	Rank among earned results
is_amazons_choice	bool	Amazon's Choice badge
best_seller	bool	Best Seller badge
sales_volume	str	For example 10K+ bought in past month

page = bee.search("fitness tracker", sort_by="featured")

AmazonScraper.featured(page)   # sponsored, ordered by ad slot
AmazonScraper.organic(page)    # earned, ordered by organic rank
AmazonScraper.badged(page)     # Amazon's Choice or Best Seller
AmazonScraper.facets(page)     # the refinements object

Two live runs on the same query returned 7 featured of 24 products, then 9 of 26. Paid density moves between requests, so measure it rather than assuming a fixed ratio.

is_prime came back False on every row of both pages, so verify it against your own target category before building a Prime filter on it.

---

Credit cost

Measured from spb-cost response headers, not quoted from a pricing page. This client stores the value on bee.last_cost after every call.

Call	Credits
Search, default light request	5 per page
Search, light_request=False	15 per page
Product, default	5
Product, light_request=False	15
Pricing	5
Any screenshot	15
Rejected request	0
usage()	0

Light requests skip the browser. They were sufficient for search, product and pricing on every call made here. Turn them off when you need review text or other content that appears only after JavaScript runs.

Failed requests are retried inside the API for up to 30 seconds, so set client timeouts above that. This client defaults to 60 seconds.

Plan tiers: ScrapingBee pricing.

---

Scope

Public Amazon listing and product pages. Nothing in this package signs in, and scraping under login credentials is prohibited by ScrapingBee's terms of service.

Related features

AI web scraping, data extraction rules, Amazon feature page, screenshots, markdown scraper, Make integration, n8n integration, Zapier integration.

The no code version of this workflow, with the Make scenario and the extraction rules, is at github.com/ScrapingBee/nocode-amazon.

License

MIT