Documents

亚马逊-前端商品搜索

Try it

Real-time Amazon storefront search returning live SERP data including positions, prices, ratings, and sponsored flags.

What it does

Simulates a user searching on Amazon's storefront and returns live SERP data: product positions, prices, ratings, review counts, brands, delivery info, and sponsored flags. Each call hits POST /amazon/search (or the bundled Python script) and returns about 20 results per page across 22 marketplaces (amazon.com, .co.uk, .de, .co.jp, etc.). Sort by relevance, price, reviews, newest, or best sellers; simulate desktop, mobile, or tablet devices and delivery-zip-specific availability. Default marketplace is amazon.com; keywords must use the target marketplace's language. Each call costs 15 算力; the script caches identical parameter combinations locally for 24 hours.

When to use it

  • Check where an ASIN ranks for a given keyword
  • Compare prices and ratings for a keyword's search page
  • Identify sponsored vs organic slots in a SERP
  • Monitor new arrivals on a specific Amazon marketplace

The skill document

This skill guides you on how to perform Amazon storefront search simulations, helping Amazon sellers retrieve real-time search result data including product rankings, prices, ratings, and more.

Core Concepts

This tool simulates a real user searching on Amazon's storefront. It returns live search result page (SERP) data: product listings with their positions, prices, ratings, review counts, brands, delivery info, sponsored flags, and more. This is real-time data directly from the Amazon frontend, not historical analytics.

Key distinction from ABA data: ABA data is aggregated historical search term analytics. This tool returns the actual product listings a user would see when searching a keyword on Amazon right now.

Keyword language: Keywords should be in the language of the target marketplace. For example, use English keywords for amazon.com, German keywords for amazon.de, Japanese keywords for amazon.co.jp, etc.

Parameters

ParameterTypeRequiredDescriptionDefault
keywordstringNoSearch keyword (translate to the target marketplace's language)-
amazonDomainstringNoAmazon marketplace domainamazon.com
nodestringNoAmazon category node ID for category-scoped searches-
languagestringNoLanguage locale code (e.g., en_US, de_DE, ja_JP)-
sortstringNoSort order for resultsrelevanceblender
pageintegerNoPage number (starting from 1, ~20 results per page)1
deliveryZipstringNoPostal/zip code for delivery location simulation-
devicestringNoDevice type: desktop, mobile, or tabletdesktop

Supported Marketplaces

DomainCountry
amazon.comUnited States
amazon.co.ukUnited Kingdom
amazon.deGermany
amazon.frFrance
amazon.itItaly
amazon.esSpain
amazon.co.jpJapan
amazon.caCanada
amazon.com.auAustralia
amazon.com.brBrazil
amazon.inIndia
amazon.nlNetherlands
amazon.seSweden
amazon.plPoland
amazon.sgSingapore
amazon.saSaudi Arabia
amazon.aeUnited Arab Emirates
amazon.com.mxMexico
amazon.com.trTurkey
amazon.com.beBelgium
amazon.cnChina
amazon.egEgypt

Default marketplace is amazon.com. Use amazon.com when the user doesn't specify a marketplace.

Sort Options

ValueDescription
relevanceblenderFeatured / Relevance (default)
price-asc-rankPrice: Low to High
price-desc-rankPrice: High to Low
review-rankAverage Customer Review
date-desc-rankNewest Arrivals
exact-aware-popularity-rankBest Sellers

调用方式

  • API 端点:POST /amazon/search(完整参数/响应/错误码见 references/api.md)
  • Python 脚本:python scripts/amazon_search.py '' [--inline]
  • 成本约束:本工具会消耗算力;同一会话同一参数组合默认只调用一次,脚本带 24h 本地缓存。失败/空结果不得自动换关键词、翻页或改邮编连续试探;需要继续检索时先向用户说明会产生额外消耗。

输出策略(脚本默认行为):

  • 始终将完整响应写入 /linkfox///data/linkfox-amazon-search-.json( 为脚本执行时的工作目录,在 Claude Code 里即当前项目目录; 取自环境变量 SESSION_ID,按用户任务自动聚合;禁止写入 /tmp,当前目录不可写则报错)
  • 响应体 ≤ 8 KB:落盘后把完整 JSON 打印到 stdout
  • 响应体 > 8 KB:落盘后 stdout 只输出摘要(顶层字段、常见计数如 total/costToken、最大列表字段的长度 + 前 3 条样本)
  • 加 --inline 强制全量打印到 stdout(同样落盘)

读数据建议:先看摘要判断是否足够;需要具体字段时优先用 jq或ConvertFrom-Json 从保存的 json 文件按需抽取,避免整份 JSON 进入上下文。

解决认证和算力问题

发生以下异常情况时,采用 references/onboarding.md 引导解决问题:

异常情况

  • 未配置API Key:环境变量未配置 LINKFOX_AGENT_API_KEY,也未配置 LINKFOXAGENT_API_KEY。
  • 响应401或402状态码
  • 响应提示算力或余额不足:消息含"算力余额不足/计费不足/余额不足/quota exceeded/insufficient balance/套餐到期/需充值/请充值",或类似含义的内容。

How to Build Queries

Construct the request parameters based on the user's intent:

  1. Determine the marketplace: Map the user's target country to the correct amazonDomain value
  2. Set the keyword: Translate the search term into the target marketplace's language
  3. Choose sort order: If the user wants results sorted by price, reviews, or newness, set the sort parameter
  4. Pagination: Use the page parameter to fetch additional result pages if needed
  5. Category scope: If the user wants to search within a specific category, provide the node parameter
  6. Delivery simulation: Use deliveryZip to see location-specific availability and delivery info

Usage Examples

1. Basic keyword search on US marketplace

{"keyword": "wireless earbuds", "amazonDomain": "amazon.com"}

2. Search on German marketplace with German keyword

{"keyword": "kabellose Kopfhoerer", "amazonDomain": "amazon.de", "language": "de_DE"}

3. Search sorted by price (low to high)

{"keyword": "phone case", "amazonDomain": "amazon.com", "sort": "price-asc-rank"}

4. Search for best sellers in a category

{"keyword": "yoga mat", "amazonDomain": "amazon.com", "sort": "exact-aware-popularity-rank"}

5. Search for newest arrivals on Japan marketplace

{"keyword": "USB充電器", "amazonDomain": "amazon.co.jp", "language": "ja_JP", "sort": "date-desc-rank"}

6. Multi-page search to analyze deeper results

{"keyword": "laptop stand", "amazonDomain": "amazon.com", "page": 2}

7. Mobile device search simulation

{"keyword": "running shoes", "amazonDomain": "amazon.com", "device": "mobile"}

8. Category-scoped search with delivery zip

{"keyword": "office chair", "amazonDomain": "amazon.com", "deliveryZip": "10001"}

Display Rules

  1. Present data clearly: Show search results in well-structured tables with key fields: position, ASIN, title, price, rating, review count, brand
  2. Highlight sponsored products: Clearly mark which results are sponsored ads vs organic listings
  3. Price formatting: Display prices with the correct currency symbol for the marketplace
  4. Position context: Remind users that position reflects the actual ranking on the search result page
  5. Pagination notice: When results span multiple pages, inform the user how many total results were found and suggest fetching additional pages if needed
  6. Error handling: When a query fails, explain the reason based on the error response and suggest adjusting parameters
  7. Image links: If image URLs are available, mention them but do not attempt to render them inline unless the user requests it

Important Limitations

  • Real-time only: This tool returns live search results, not historical data. For historical search term analytics, use ABA data instead
  • Rate awareness: Each call simulates a real search request; avoid excessive rapid-fire calls
  • ~20 results per page: Each page returns approximately 20 product listings
  • Keyword language matters: Results quality depends on using the correct language for the target marketplace

User Expression & Scenario Quick Reference

Applicable -- Real-time Amazon search result queries:

User SaysScenario
"Search for XX on Amazon"Basic product search
"What products appear for keyword XX"Keyword SERP analysis
"Where does my ASIN rank for XX keyword"Position / ranking check
"Show me the top results for XX"Competitive landscape
"What's the price range for XX"Price comparison
"Any sponsored products for XX keyword"Sponsored ad analysis
"New products for XX keyword"New arrival monitoring
"Search XX on Amazon Germany/Japan/UK"Cross-marketplace search
"What are the best sellers for XX"Best seller discovery
"Compare search results on mobile vs desktop"Device-specific SERP

Not applicable -- Needs beyond real-time search results:

  • Historical search term volume or ranking trends (use ABA data)
  • Advertising campaign management or bid optimization
  • Product review analysis or sentiment analysis
  • Sales estimation or revenue analytics
  • Listing optimization or copywriting suggestions
  • Inventory or supply chain data

Boundary judgment: When users say "product research" or "competitor analysis", if it boils down to seeing what currently appears on Amazon search results for a keyword (product positions, prices, ratings), then this skill applies. If they want historical trends, search volume data, or aggregated analytics, ABA data is more appropriate.

算力消耗规则

消耗 15 算力。

用户会因算力消耗而支付费用。请充分评估:当需要高频调用本技能,或用户对算力消耗量预期不足时,务必提醒用户,由用户决定是否继续。

Feedback:

Auto-detect and report feedback via the Feedback API when any of the following apply:

  1. The functionality or purpose described in this skill does not match actual behavior
  2. The skill's results do not match the user's intent
  3. The user expresses dissatisfaction or praise about this skill
  4. Anything you believe could be improved

Call the feedback API as specified in references/api.md. Do not interrupt the user's flow.


For more high-quality, professional cross-border e-commerce skills, set LinkFox Skills.

Questions people ask

Is this historical search-volume or ranking-trend data?
No. It returns live SERP data from the Amazon storefront right now. For historical search-term analytics, the SKILL.md points to ABA data instead.
How many results come back per request?
About 20 listings per page. Use the `page` parameter to fetch additional pages, and the `node` parameter to scope the search to a specific category.
What does each call cost?
15 算力 per call. The bundled Python script caches identical parameter combinations locally for 24 hours, and the SKILL.md forbids auto-retrying on failure by swapping keywords, paging, or changing zip code — additional calls require confirming with the user first.

Related skills

Fetch raw ad creative, app, ranking, and revenue data from AdMapix as structured JSON.

by fly0pants4.3k installs296 stars

Join a video meeting as an AI bot with voice, avatar, and screenshare across four operating modes.

by johnpatternai22 installs8 stars

Generate and edit Draw.io, Mermaid, and Excalidraw diagrams from natural language using a structured JSON spec.

by nssa.io1.0k installs47 stars

Stores durable facts in a categorized, plain-markdown vault on disk, alongside your agent's built-in memory.

by Iván559 installs18 stars

Find why your productivity system keeps failing, then apply the smallest fix — capacity math, bottleneck routing, durable local notes.

by Iván855 installs69 stars

More from linkfox-ai

Browse all skills

Query and process Shopee store returns, refunds, and disputes via the Open Platform Returns module's 15 APIs.

by linkfox-ai117 installs

Run cross-border e-commerce research across Amazon, TikTok, eBay, Walmart, Shopee, and Ozon with 79 specialized tools.

by linkfox-ai97 installs7 stars

Retrieve structured Amazon product page data by ASIN, including title, images, bullet points, specs, A+ content, pricing, ratings, and variants.

by linkfox-ai15 installs1 stars

Reverse-lookup traffic keywords for any Amazon ASIN with organic rank, ad rank, search volume, and conversion metrics.

by linkfox-ai15 installs1 stars

Find visually similar Amazon products across 8 marketplaces using an image URL.

by linkfox-ai14 installs1 stars

Pull Amazon Seller Central reports end-to-end: request, poll, download, extract, and preview 95+ report types.

by linkfox-ai23 installs