How to Access Amazon Product Data via Shadow APIs

Amazon product data is some of the most valuable and most difficult to access on the web. Pricing, reviews, availability, seller information, product specifications — this data drives pricing intelligence tools, market research platforms, competitive analysis, and e-commerce automation. But Amazon has spent years building one of the most sophisticated anti-scraping systems on the internet.

If you need Amazon data for an AI agent, price monitoring tool, or product research pipeline, the traditional approaches are expensive, unreliable, or both.

The Problem with Amazon Data Access

Amazon's data access landscape is dominated by pain:

• Product Advertising API (PA-API): Amazon's official API requires an active Associates account with qualifying sales. Rate limited to 1 request/second. Limited to search and product detail — no review data, no seller data, no price history. Requires a revenue-generating affiliate account to maintain access.
• HTML Scraping: Amazon runs one of the most aggressive anti-bot systems in existence. CAPTCHAs, IP blocking, browser fingerprinting, behavioral analysis, and request pattern detection. Even sophisticated rotating proxy setups get blocked within hours.
• Third-party scraping services: Oxylabs, Bright Data, ScrapingBee offer Amazon-specific products at $100-500/month. Data freshness varies. Complex pricing models. You're paying for proxy infrastructure to fight Amazon's anti-bot.
• Amazon SP-API: For registered sellers only. Requires a professional seller account ($39.99/month) and API approval. Covers only your own listings and orders.

For AI agents that need to check a product's price, read reviews, or compare options, there's no good programmatic path.

Shadow APIs: The Alternative

Every time you visit amazon.com, your browser makes API calls behind the scenes. These internal endpoints return clean JSON data — the same data that renders the product pages, search results, and review sections you see.

Amazon's frontend makes calls to internal services for dynamic content: pricing updates, review loading, availability checks, recommendation carousels, and search results. These endpoints return structured JSON that's far cleaner than parsing Amazon's notoriously complex HTML.

Unbrowse captures these shadow APIs automatically from real browsing sessions.

What Unbrowse Discovers on Amazon

Browsing Amazon through Unbrowse reveals internal API endpoints that power the shopping experience:

• Search Results: GET /s/query?k={keywords}&ref=... — product search results with structured data including ASIN, title, price, rating, review count, Prime eligibility, and thumbnail URLs. Supports filters for price range, brand, rating, and department.
• Product Detail: GET /dp/{asin}/ref=... combined with POST /gp/product/features/glance/v2/ — product details including all variations (size, color), current price, list price, availability status, delivery estimates, and seller information.
• Reviews: GET /gp/customer-reviews/widgets/average-customer-review/popover?asin={asin} and GET /reviews/widgets/cr-carousel/data?asin={asin} — customer reviews with star breakdowns, top reviews, verified purchase status, and helpfulness votes.
• Price and Availability: POST /gp/aod/ajax?asin={asin} — real-time pricing from all sellers including new/used prices, shipping costs, and seller ratings. This is the endpoint behind the "Other Sellers on Amazon" section.
• Recommendations: GET /gp/product/ajax/ref=...?asin={asin}&experienceId=... — "Frequently bought together," "Customers also viewed," and sponsored product recommendations.

These endpoints return the live data that Amazon's frontend renders — current prices, real-time availability, and up-to-date reviews.

How It Works

npm install -g unbrowse
unbrowse resolve "search for wireless earbuds under 50 dollars" --url https://amazon.com

The process:

1. Cookie Injection: Unbrowse extracts your Amazon session cookies from your local browser. Your logged-in session is reused, which means you see the same prices and availability as you would browsing normally (including Prime pricing if you're a member).
2. Browse: Kuri opens Amazon with your real session and anti-detection extensions. Amazon sees a legitimate browser session.
3. Capture: Every internal API call is intercepted — search queries, product detail fetches, review loads, price checks. Both HAR recording and the fetch/XHR interceptor ensure nothing is missed.
4. Index: Endpoints are analyzed for URL templates (ASIN patterns, search parameters), auth requirements (session cookies, CSRF tokens), and response schemas.
5. Cache and Execute: Indexed routes are stored. Future requests call Amazon's internal APIs directly with your session credentials.

Performance

Metric	Browser Automation (Playwright)	Scraping Service (Oxylabs)	Unbrowse (cached)
Speed	~4,500ms	~2,000ms	<120ms
Tokens consumed	~20,000	N/A	~350
Cost per query	$0.53	$0.002-0.01	$0.005
Anti-bot risk	Very high	Medium (proxy rotation)	None
Data freshness	Real-time	Minutes to hours	Real-time
Monthly cost	Compute + proxies	$100-500	Free (open source)

The biggest win is reliability. Amazon's anti-bot system breaks Playwright scrapers constantly. Unbrowse uses your real session, so there's nothing to detect or block.

When to Use This Approach

Price monitoring: Track product prices across categories for competitive intelligence, deal alerts, or pricing optimization. The price/availability endpoint provides real-time data from all sellers without rate limits.

Product research for AI agents: Your agent needs to compare products, check reviews, or find the best option for a user's request. Instead of parsing complex HTML, Unbrowse gives structured JSON with all product attributes.

Review analysis: Pull review data for sentiment analysis, product quality assessment, or competitive research. The review endpoints provide structured data with ratings, verification status, and helpfulness metrics.

E-commerce intelligence: Monitor new product launches, bestseller rankings, and category trends. Search and recommendation endpoints provide the same discovery data Amazon's algorithms surface to shoppers.

Getting Started

# Install Unbrowse globally
npm install -g unbrowse

# Run initial setup
unbrowse setup

# Make sure you're logged into Amazon in Chrome, then:
unbrowse resolve "search for mechanical keyboards" --url https://amazon.com

# Cached routes work for any Amazon query
unbrowse resolve "find reviews for ASIN B09V3KXJPB" --url https://amazon.com

Unbrowse is open source and published on arXiv. It works as an MCP server for AI agents.

FAQ

Is this legal? Unbrowse uses your authenticated browser session and accesses the same APIs your browser calls when you shop on Amazon. You're accessing data visible to you as a logged-in customer.

How is this different from scraping? Scraping parses Amazon's complex HTML, which changes frequently and is designed to be difficult to parse. Unbrowse calls the actual internal APIs that Amazon's frontend uses — structured JSON responses that are stable and complete.

Does this work with Prime pricing? Yes. Because Unbrowse uses your real session cookies, it sees the same prices you see — including Prime member pricing, Subscribe & Save discounts, and personalized deals.

What about Amazon's anti-bot detection? Unbrowse runs a real browser with your real credentials. There's no bot to detect. The API calls use your actual session at human-like rates. Amazon's anti-bot systems target automated tools that forge sessions or use proxy networks.

Can I monitor prices over time? Yes. Once Unbrowse has cached the price/availability endpoint, you can poll it periodically. The endpoint returns real-time data, so you get current prices on every call.