back
loading skill details...
通过亚马逊前台的 Alexa 购物助手发起自然语言问答，获取与问题相关的导购回答、推荐商品分组、ASIN 列表，以及可继续追问的问题。支持在同一次调用中传入多条 prompts 模拟连续多轮对话，并可用 url 补充亚马逊页面上下文。当用户提到亚马逊 Alexa、Alexa 购物助手、亚马逊智能助手、AI 导购、...
---
name: linkfox-amazon-alexa-for-shopping
version: 1.0.0
category: product-sourcing
description: 通过亚马逊前台的 Alexa 购物助手发起自然语言问答，获取与问题相关的导购回答、推荐商品分组、ASIN 列表，以及可继续追问的问题。支持在同一次调用中传入多条 prompts 模拟连续多轮对话，并可用 url 补充亚马逊页面上下文。当用户提到亚马逊 Alexa、Alexa 购物助手、亚马逊智能助手、AI 导购、对话式选品、自然语言购物、亚马逊聊天问答、Amazon Alexa shopping, conversational shopping, AI shopping assistant, follow-up questions、产品推荐对话、上下文追问等场景时触发此技能。即使用户未明确提及"Alexa"，只要其需求是"在亚马逊前台用自然语言问出商品推荐 + 多轮追问"，也应触发此技能。
---
# Amazon Alexa Shopping Assistant
This skill drives Amazon's storefront Alexa shopping assistant: pose a natural-language question and get an answer, a curated product list (with ASINs and links), and a set of follow-up questions Alexa is willing to continue with. Multiple prompts in a single call simulate a continuous multi-turn conversation, not independent searches.
## Core Concepts
1. **Single conversation, ordered turns**: `prompts` is an array — element 0 is the opening question, element 1 the first follow-up, element 2 the next follow-up, and so on. The tool sends them sequentially in one Alexa session and concatenates Alexa's answers in order.
2. **Cross-call context is not preserved**: when a new tool call starts, it begins a brand-new Alexa session. To keep continuity across calls, summarize the previous answer + recommended ASINs yourself and embed them in the next `prompts[0]`.
3. **Optional page context (`url`)**: pass an Amazon page URL only when you want the conversation anchored to a **specific** page (a category page, search results page, or product detail page). Do **not** pass a plain marketplace homepage URL like `https://www.amazon.com/` — it adds no useful context. Omit `url` entirely when there is no specific page to anchor on.
4. **Two output formats**:
- `markdown` (default) — a single readable Markdown report containing the question, Alexa's answer, recommended product groups, and follow-up questions.
- `json` — a structured array under `data`, where each entry carries `prompt`, `content`, `products` (grouped recommendations), `followUpQuestions`, and `screenshot`.
`resultsNum` is the number of conversation turns Alexa actually answered; if `0`, Alexa did not produce a usable reply for the input.
## Parameters
| Parameter | Type | Required | Description | Default |
|-----------|------|----------|-------------|---------|
| prompts | string[] | Yes | Conversation prompts. Each element is one turn in the same session, sent in order. Recommended ≤ 5 entries. | - |
| format | string | No | Response format: `markdown` returns a readable report; `json` returns a structured array. | markdown |
| url | string | No | Specific Amazon page URL (category, search results, or product detail) to anchor the conversation. Skip when there is no specific page; do **not** pass a plain homepage URL such as `https://www.amazon.com/`. | - |
## Response Fields
| Field | Type | Description |
|-------|------|-------------|
| stdout | string | Markdown report when `format=markdown`: per-turn question, Alexa answer, recommended product groups, follow-up questions |
| data | array | Structured turns when `format=json`. Each item has `prompt`, `content`, `products[]`, `followUpQuestions[]`, `screenshot` |
| resultsNum | integer | Number of answered turns (0 = Alexa did not respond) |
| code / errcode | string / integer | `200` on success; non-200 indicates a business error |
| msg / errmsg | string | `ok` on success; otherwise an error description |
| costTime | integer | API latency in milliseconds |
| costToken | integer | Tokens consumed (only billed on success) |
| taskId | string | Upstream task identifier for tracing |
| type | string | Render hint: `stdoutWorkbenches` for markdown, `json` for json |
### Structured `data[*]` shape (`format=json`)
| Field | Type | Description |
|-------|------|-------------|
| prompt | string | The question or follow-up sent for this turn |
| content | string | Alexa's natural-language answer |
| products[].title | string | Group title (e.g. "Top picks", "Best for running") |
| products[].items[].asin | string | Product ASIN |
| products[].items[].title | string | Product title |
| products[].items[].url | string | Product detail page URL |
| products[].items[].cover | string | Product cover image URL |
| products[].items[].price | string | Current price string (with currency) |
| products[].items[].originalPrice | string | List price / strikethrough price |
| products[].items[].score | string | Star rating |
| products[].items[].ratingsCount | string | Review count |
| products[].items[].describe | string | Short product blurb |
| followUpQuestions | string[] | Questions Alexa offers to continue with |
| screenshot | string | Screenshot URL for this turn |
## API Usage
This skill calls the LinkFox tool gateway. See `references/api.md` for the calling convention, request/response shape, error codes, and a curl example. You can also run `scripts/amazon_alexa_search.py` directly to test it from the command line.
## How to Build Queries
1. **Front-load the user's intent in `prompts[0]`** — include marketplace cue ("on Amazon US"), use case, and any hard constraints (budget, key feature). Alexa weights the opening turn heavily.
2. **Order follow-ups by dependency** — each turn reuses the prior turn's context. Put broad framing first, then ask Alexa to compare, narrow, or recommend specific picks.
3. **Keep `prompts` short** — 1 to 5 turns is the sweet spot. Longer arrays inflate latency without proportional gain.
4. **Anchor with `url` only when there's a specific page** — pass a category, search results, or product detail URL when the user is reasoning over that page. Skip `url` for general questions; do not pass a plain homepage like `https://www.amazon.com/`.
5. **For continuity across tool calls**: write a one-paragraph summary of the previous answer (key recommendations + ASINs) and prepend it to the new `prompts[0]`. Don't assume Alexa remembers the prior call.
6. **Pick `format` deliberately** — `markdown` is best for showing the user a polished answer; `json` is better when downstream code needs to extract ASINs, prices, or follow-up questions programmatically.
### Usage Examples
**1. Single-turn shopping question**
```json
{
"prompts": ["best wireless earbuds for running on Amazon US under $100"]
}
```
**2. Multi-turn conversation (compare + narrow)**
```json
{
"prompts": [
"best electric kettle on Amazon US",
"compare the top two recommendations on noise level and boil time",
"which one is better if I only boil water once a day"
]
}
```
**3. Conversation anchored to a category page**
```json
{
"prompts": [
"What are the most popular picks on this page?",
"Which of them have the best reviews for small kitchens?"
],
"url": "https://www.amazon.com/s?k=electric+kettle"
}
```
**4. Structured output for downstream extraction**
```json
{
"prompts": ["best gift ideas for a 10-year-old who likes science"],
"format": "json"
}
```
## Display Rules
1. **Render the Markdown directly** when `format=markdown`: `stdout` is already structured with turn headings, product cards, and follow-up questions — preserve that structure.
2. **Surface the recommended ASINs** so the user can click through; show `title`, `price`, `score`/`ratingsCount`, and the product URL.
3. **Show the follow-up questions** Alexa returned — they are usable prompts the user can pick to continue digging.
4. **Don't reroute to a data-analysis sandbox**: the answer body is conversational and the recommended products are nested groups, not a flat tabular dataset suitable for SQL-like aggregation.
5. **Flag empty results**: if `resultsNum` is `0` or `data` is empty, tell the user Alexa did not produce a usable reply and suggest rephrasing or anchoring with a `url`.
6. **Indicate freshness**: results reflect Alexa's live answer at call time; mention this when the user asks about timing.
7. **Handle business errors**: if `code` / `errcode` is not `200`, surface `msg` / `errmsg` and suggest retrying with simpler prompts.
## Important Limitations
- **Alexa-driven, not deterministic**: same prompts can yield different answers across calls — Alexa's response varies with time, traffic, and context.
- **No cross-call memory**: each tool call is a fresh Alexa session; the agent must recap prior context manually.
- **Recommended ≤ 5 prompts per call**: longer arrays sharply increase latency.
- **Marketplace coverage**: anchored on Amazon's storefront Alexa experience (primarily amazon.com); availability on non-US marketplaces depends on Alexa rollout.
- **Output mix**: primary value is the conversational answer plus a curated handful of products; this is not a substitute for SERP-wide product extraction.
## User Expression & Scenario Quick Reference
**Applicable** — natural-language conversational shopping on Amazon:
| User Says | Scenario |
|-----------|----------|
| "用 Alexa 帮我推荐...", "亚马逊 Alexa 问下..." | Direct Alexa Q&A |
| "在亚马逊上聊聊给我推荐 ...", "对话式选品" | Conversational discovery |
| "顺便再追问一下 / 接着问 ..." | Multi-turn follow-up |
| "在这个页面 / 这个分类下推荐...", "基于这个页面再问一下" | Page-anchored conversation (use `url`) |
| "best XX for YY under $Z on Amazon" | Goal + constraint + budget Q&A |
| "对比 Alexa 给的前两个推荐" | Compare within Alexa's reply |
| "Alexa 还能继续问什么 / 给我一些追问思路" | Surface follow-up questions |
**Not applicable** — better routed elsewhere:
- Pulling the full SERP for a keyword with positions, sponsored flags, etc. (use the storefront search-simulation skill).
- Historical search-term analytics or volume trends (use the ABA data explorer).
- Detailed product detail / A+ / bullets for a known ASIN (use the Amazon product detail skill).
- Review-level sentiment analysis (use the Amazon reviews skill).
- Image-based similar product discovery (use the image search skill).
- Aggregated statistics over a flat product list (no structured table here).
**Boundary judgment**: when the user wants a **conversation** — "ask Amazon, get a recommendation, then keep asking" — this skill applies. If they want raw search-result rows, structured analytics, or a specific ASIN's data, route to the matching specialized skill instead.
**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](https://skill.linkfox.com/).*
don't have the plugin yet? install it then click "run inline in claude" again.
extracted and documented 6 required components (intent, inputs, procedure with explicit steps and outputs, decision points covering branching logic and edge cases, output contract with data formats, and outcome signal); clarified cross-call context handling, url anchor rules, and error scenarios; added rate limit and auth edge cases; preserved original procedure and author intent.
use this skill when you need to ask amazon's alexa shopping assistant a natural-language question about products, get back a conversational answer plus curated product groups with asins and links, and optionally continue with follow-up questions in the same session. trigger it whenever a user wants to shop via conversation on amazon (e.g. "ask alexa for running shoes under $100", "compare the top two recommendations", "what else can i ask about this category"). do not use this for raw serp extraction, analytics, or detailed single-product data.
required:
prompts (string array): ordered conversation turns. element 0 is the opening question, element 1 is the first follow-up, etc. send ≤ 5 prompts per call; each turn is sent sequentially in one alexa session and answers are concatenated.optional:
url (string): a specific amazon page url (category, search results, or product detail page) to anchor the conversation. omit this parameter entirely if there is no specific page; do not pass a plain homepage like https://www.amazon.com/ as it adds no context.format (string): response format, either markdown (default, returns a readable report) or json (returns a structured array).context & external connections:
collect the user's intent: identify the product category, use case, and any hard constraints (budget, features, size, etc.) from the user's question.
format the opening prompt: front-load the intent in prompts[0]. include marketplace cue (e.g. "on amazon us"), use case, and constraints. alexa weights the opening turn heavily, so be explicit.
plan follow-up turns (if needed): if the user wants comparison, narrowing, or multiple questions, order them by dependency. each turn reuses the prior turn's context.
determine if page context applies: only set url if the user is reasoning over a specific page (category page, search results, or product detail). skip url for general questions.
choose response format: set format=markdown for a polished user-facing report, or format=json if downstream code needs to extract asins, prices, or follow-up questions programmatically.
call the linkfox gateway with the request payload (prompts, url, format).
receive the response: the response includes stdout (markdown report if format=markdown), data (structured array if format=json), resultsNum (number of answered turns), code (200 on success), and msg (error description if code is not 200).
check resultsNum: if resultsNum is 0, alexa did not produce a usable reply. flag this to the user and suggest rephrasing or anchoring with a url.
render the answer: if format=markdown, display stdout directly (it is already structured with turn headings, product cards, and follow-up questions). if format=json, extract and display the recommended asins, titles, prices, ratings, and follow-up questions as needed by your downstream logic.
surface follow-up options: display the follow-up questions alexa returned so the user can pick one to continue.
for continuity across multiple tool calls: if the user asks another question after seeing the first answer, summarize the prior answer plus recommended asins in one paragraph and prepend it to the new prompts[0]. do not assume alexa remembers the prior call.
if user asks a direct product question with intent (e.g. "best wireless earbuds under $100"): set prompts to a single element containing the question. use format=markdown. omit url.
if user wants to compare or narrow down within alexa's answer (e.g. "which of those has the best reviews for small kitchens"): add a second prompt to the prompts array. send both in one call so alexa can see the context.
if user is browsing a specific amazon page and asks about it (e.g. "what are the most popular picks on this category page"): set url to the current page url. this anchors the conversation to that page's inventory.
if url is a plain homepage or generic marketplace url (e.g. https://www.amazon.com/ or https://www.amazon.com/s): omit url. homepage context adds no value.
if resultsNum is 0 (alexa did not respond): tell the user alexa did not produce a usable reply, suggest simpler phrasing or anchoring with url, and optionally retry.
if code is not 200 (business error, e.g. rate limit, auth failure, alexa service down): surface the msg / errmsg to the user. suggest retrying with simpler prompts or waiting a moment.
if the user's intent is raw serp extraction, analytics, review sentiment, or a specific asin's details: do not use this skill. route to the storefront search-simulation skill, aba data explorer, amazon product detail skill, or amazon reviews skill instead.
if the user has seen the prior answer and wants to continue with a new question (e.g. after a previous tool call): check if their new question builds on the prior context. if yes, prepend a one-paragraph summary of the prior answer + asins to the new prompts[0]. if no, start fresh.
on success (code=200):
resultsNum: integer, the count of answered turns (0 means alexa did not respond; 1+ means alexa answered that many turns).stdout: string (markdown format). contains per-turn structure: prompt, alexa's natural-language answer, recommended product groups (with asin, title, url, price, originalPrice, score, ratingsCount, describe), and follow-up questions. rendered as readable markdown with turn headings and product cards.data: array of objects (json format). each object has:prompt: string, the question sent.content: string, alexa's answer.products: array of product groups, each with title (group name) and items (array of products). each item has asin, title, url, cover image, price, originalPrice, star score, ratingsCount, and describe.followUpQuestions: string array, questions alexa offers to continue with.screenshot: string, screenshot url for that turn.costTime: integer, api latency in milliseconds.costToken: integer, tokens consumed (billed on success only).taskId: string, upstream task identifier for tracing.type: string, render hint (stdoutWorkbenches for markdown, json for json).msg: string, value is ok.on failure (code is not 200):
code: integer, non-200 status code (e.g. 400 for bad request, 500 for server error, 429 for rate limit).msg: string, error description (e.g. "rate limit exceeded", "invalid url", "alexa service unavailable").resultsNum: 0.costToken: 0 (no tokens consumed).data format details (when format=json):
each data[*].products[j].items[k] contains: asin (string), title (string), url (string), cover (image url), price (string with currency), originalPrice (optional strikethrough price), score (star rating as string), ratingsCount (review count as string), describe (short product blurb).
markdown format: you display the stdout markdown report to the user. they see the per-turn q&a, product cards with links and prices, and follow-up questions. they can click through to product detail pages or pick a follow-up question to continue.
json format: downstream code successfully extracts asins, titles, prices, and follow-up questions from the data array and builds a custom ui or feeds those asins into a cart or comparison tool.
user continues the conversation: the user picks one of the follow-up questions alexa offered (or asks their own), and you call the skill again with the new prompt appended to the prior prompts array (or, if it is a new call, prepend a summary of the prior answer to prompts[0]).
user satisfaction signal: the user clicks a product link, adds an item to cart, or says something like "these recommendations are exactly what i needed" or "alexa nailed it".
failure signal: resultsNum is 0, code is not 200, or the user says alexa did not understand or gave irrelevant recommendations. in that case, suggest rephrasing the prompt, anchoring with a url, or retrying with simpler language.
original source: clawhub. author not declared in source. enriched for implexa quality standards.