amazon-shopping

name	amazon-shopping
version	1.1.0
description	Search Amazon, extract product data, and present ranked recommendations. Use when user asks to shop on Amazon, find products, compare items, or research purchases. Prioritizes review count over rating.
allowed-tools	mcp__chrome-devtools__*

Amazon Shopping

Search Amazon and recommend products based on user preferences.

---

MANDATORY STEP 0 - SELECT AMAZON DOMAIN

Before anything else, determine which Amazon regional site to use. Prices, availability, currency, and shipping all depend on the domain — searching the wrong one wastes the whole interaction.

Detection rules (skip the question if any apply)

If the user's request already specifies the marketplace, use it directly without asking:

• Explicit domain mentioned: "amazon.fr", "amazon.de", "amazon.co.uk", etc.
• Country named: "Amazon France", "on the German Amazon", "UK Amazon"
• Currency hint: prices in €, £, ¥, C$, etc.
• Parameter passed by the caller (e.g. domain=amazon.it)

Map these to the corresponding domain from the supported list below and proceed.

When to ask

If the request is ambiguous (e.g. "find me headphones on Amazon", "shop for a coffee maker"), use AskUserQuestion with this closed list of supported domains:

Label	Domain
United States (.com)	amazon.com
France (.fr)	amazon.fr
Germany (.de)	amazon.de
United Kingdom (.co.uk)	amazon.co.uk
Italy (.it)	amazon.it
Spain (.es)	amazon.es
Canada (.ca)	amazon.ca
Japan (.co.jp)	amazon.co.jp
Australia (.com.au)	amazon.com.au
Mexico (.com.mx)	amazon.com.mx
Brazil (.com.br)	amazon.com.br
India (.in)	amazon.in
Netherlands (.nl)	amazon.nl

Store the chosen domain as <AMAZON_DOMAIN> and use it for all subsequent URLs in this session (search, product pages, output links).

---

MANDATORY STEP 1 - REQUIREMENTS GATHERING

YOU MUST ASK CLARIFYING QUESTIONS BEFORE ANY BROWSER AUTOMATION.

DO NOT proceed to search until you understand:

• Budget: Price range they're comfortable with (in the local currency of <AMAZON_DOMAIN>)
• Usage: Who/what it's for (personal, gift, professional)
• Deal-breakers: Features they must have or avoid

Use AskUserQuestion to gather this information. Only after receiving answers should you proceed to Step 2.

Example questions for any product:

• Budget range (under $25, $25-50, $50-100, $100+)
• Primary use case (personal, gift, professional)
• Key preferences (brand, features, quality vs value)

---

Quick Start (After Requirements Gathered)

1. Gather requirements: Ask about budget, usage, deal-breakers → DONE in mandatory step above
2. Search Amazon: Navigate to search results and capture snapshot
3. Extract data: Identify products with ASINs, prices, ratings, review counts
4. Verify: Visit product pages to confirm accuracy
5. Present recommendations: Ranked by user's criteria

Prerequisites

• Chrome DevTools MCP server connected
• Internet access to Amazon.com

Search Workflow

Step 1: Navigate to Amazon Search Results

Go directly to the search URL (simplest approach):

navigate_page type="url" url="https://www.<AMAZON_DOMAIN>/s?k=<encoded search query>"

Alternative - Interactive search (if direct URL doesn't work):

1. Navigate to https://www.<AMAZON_DOMAIN>
2. take_snapshot to find the search box UID
3. fill the search box with your query
4. click the search button or press_key with key="Enter"

Step 2: Capture Results

take_snapshot

The snapshot returns the full accessibility tree with UIDs for every element. Each element includes its role, name, and URL where applicable.

Step 3: Extract Products WITH Their ASINs

CRITICAL: Extract product names AND their ASINs from THE SAME container element to prevent mismatches.

In the accessibility tree, each product is typically a listitem containing:

• A heading with the product name
• A link with the ASIN in the URL (e.g., /dp/B0XXXXXXXXX)

Read the snapshot and pair each heading with the link in the same container. Use the UID hierarchy and indentation to identify which elements belong to the same product.

WRONG - Causes ASIN/Product Mismatches:

• Extracting all ASINs separately and pairing them with product names
• Amazon pages contain ASINs for ads, sponsored products, and footer links that will mix with real results

Step 4: MANDATORY Product Page Verification

For EVERY product before presenting it to the user, you MUST verify by visiting its actual product page.

For each product:

1. navigate_page to https://www.<AMAZON_DOMAIN>/dp/<ASIN>
2. take_snapshot to read the page content
3. Verify:
   • Title matches the product you're recommending
   • Price is current from the product page (for example Buy new: $XX.XX, One-time purchase: $XX.XX, or the main price block)
   • Rating and review count are present

MANDATORY VERIFICATION RULES:

• If ASIN redirects to different product → DISCARD
• If page title doesn't match expected product → DISCARD
• If price cannot be found → DISCARD
• Only present VERIFIED products with a checkmark

NEVER extract prices from search results pages - they often don't match actual product page prices due to variants, promotions, or different sellers.

See reference/asin-extraction.md for detailed patterns.

Common Issues

Issue	Solution
CAPTCHA	Wait 60s, retry from <AMAZON_DOMAIN>
Rate limited	Wait 2-3 min
No results	Broaden search
Slow page load	Use wait_for with expected visible text, then retake take_snapshot

See reference/common-errors.md for complete troubleshooting.

Output Format

ALL presented products MUST be verified on their actual product pages.

## Amazon Shortlist - [Category] (<AMAZON_DOMAIN>)

### 1. [Product] - [LOCAL_CURRENCY][PRICE] (verified)
**ASIN**: [ASIN]
**Rating**: X.X/5 (X,XXX reviews)
**<AMAZON_DOMAIN>**: https://www.<AMAZON_DOMAIN>/dp/[ASIN]
**Why this**: [Reason]
**Key specs**: [Specs]

Do NOT present unverified products. The verified marker confirms that:

1. The ASIN link goes to the actual product (not a redirect)
2. The price is current from the product page
3. The title matches what was recommended

See reference/output-formats.md for templates.

Ranking Priority

When ratings are similar (within 0.5), prioritize review count.

Example: 4.0 with 10,000 reviews > 5.0 with 100 reviews