{"id":86711,"plugin_id":"plugin_asdk_app_6acd20c30cf08191889c13e8f7dfd318","kind":"skill","collection_source":"plugin_package","comparison_source":null,"observed_at":"2026-11-09T01:28:31.326Z","digest":"e959ec1f4be4606885375edab4ac8e8542cc63132476c89b7c1857e342ecb664","against":null,"payload":{"description":"Research Amazon products, niches and brands with AMZScout's live Amazon data. Use when the user wants to check whether a product is worth selling, analyze a niche or its competition, find product ideas, look up a listing's keywords and ads, or review a brand's Amazon catalog.\n","included_files":[],"name":"amzscout-research","skill_md_contents":"---\nname: amzscout-research\ndescription: >\n  Research Amazon products, niches and brands with AMZScout's live Amazon data.\n  Use when the user wants to check whether a product is worth selling, analyze a\n  niche or its competition, find product ideas, look up a listing's keywords and\n  ads, or review a brand's Amazon catalog.\n---\n\n# AMZScout research\n\nThis skill teaches you to orchestrate the granular AMZScout tools yourself: pick\nthe right tool(s), chain them into a workflow, and turn the raw data into a\nclear seller-facing answer. The tools come from the `amzscout` MCP server.\n\nIf the AMZScout tools are not available in this session, say so — do not\nfabricate AMZScout data.\n\n## How to use AMZScout\n\n- **The granular `amzscout_*` tools** (below) are the building blocks, and on a\n  direct MCP connection they are ALL you get. You choose the workflow, chain the\n  steps, and compose the answer yourself. This skill is about using them well.\n\n\n## Granular tools (MCP)\n\n| Tool | Args | Returns |\n|------|------|---------|\n| `amzscout_locate_asin` | `asins` (1–5) | Where an ASIN is listed: each marketplace's link, title, local price, sales. Uses 14 000 tokens of the account's data allowance per ASIN — only when the user asks where a product is sold (see Marketplaces). |\n| `amzscout_analyze_product` | `asin`, `marketplace?`, `focus?` | Full audit of ONE product: verdict, demand/trend, pricing, competition, maturity, listing quality, GO/NO-GO. |\n| `amzscout_reseller_amazon` | `asin`, `marketplace?`, `sections?` | Reseller / online-arbitrage view of ONE ASIN: buy box owner and price, FBA vs FBM offer counts, whether Amazon sells it, live offers, price levels 30/90/180/365 days, rank drops, out-of-stock share, buy box history. Call WITHOUT `sections` first — the summary usually answers; add sections (`buybox`, `competition`, `pricing`, `rank`, `availability`, `history`) only if it does not. Not for demand/revenue — that is `amzscout_analyze_product`. |\n| `amzscout_compare_products` | `asins` (2–5), `marketplace?` | Side-by-side comparison of 2–5 ASINs — which is the stronger opportunity. |\n| `amzscout_compare_niches` | `keywords` (2–5), `marketplace?`, `count?` | Head-to-head comparison of 2–5 niches/keywords — which niche is the better opportunity. |\n| `amzscout_analyze_product_set` | `asins` (2–25), `focus?`, `marketplace?` | Market overview across an explicit set of ASINs (you already have the list). |\n| `amzscout_analyze_niche` | `keyword`, `marketplace?`, `filters?`, `count?` | Niche/category overview: segments, top products, demand concentration, signals. |\n| `amzscout_search_products` | `query`, `marketplace?`, `sort?`, `count?`, `filters?` | Discovery: a list of products for a keyword (no deep analysis). |\n| `amzscout_find_by_brand` | `brand`, `marketplace?`, `sort?`, `count?`, `filters?` | Products listed under a brand on Amazon. |\n| `amzscout_get_keywords` | `asin?`, `keyword?` (one or both), `marketplace?` | Keyword / SEO / PPC data — ASIN-scope (terms a product ranks for), keyword-scope (search data around a niche term), or **both** = the product's position for exactly that keyword. |\n| `amzscout_search_knowledge` | `query`, `topK?` | AMZScout knowledge-base passages (how-to, AMZScout tool/feature info). |\n| `amzscout_usage` | — | The user's AMZScout token balance (remaining / used / limit). Free. Use for \"how many tokens do I have left\", \"my usage/balance\". |\n\n## When to use which\n\n- Specific product — ASIN (`B0XXXXXXXX`) or Amazon link → `amzscout_analyze_product`. Bare ASIN with no marketplace known in this chat → ask the user once which marketplace (see Marketplaces).\n- Reselling / online arbitrage / wholesale about a specific ASIN — \"who has the buy box\", \"how many FBA/FBM sellers\", \"does Amazon sell it\", \"is this worth reselling\", buy box price history → `amzscout_reseller_amazon`. Private-label \"should I sell something like this\" stays `amzscout_analyze_product`.\n- \"Where / on which Amazon marketplaces is B0… sold\" → `amzscout_locate_asin` (uses data allowance).\n- A niche / category / keyword phrase, or \"is X worth entering / how saturated is X\" → `amzscout_analyze_niche`.\n- Product-design / attribute question about a niche (\"what materials / sizes / packaging work for X\") → `amzscout_analyze_niche` shows what the top sellers in the niche use — ground the design advice in that.\n- 2–5 ASINs to compare head-to-head → `amzscout_compare_products`. A set of ASINs you want one combined overview of → `amzscout_analyze_product_set`.\n- 2–5 niches/keywords to compare (\"yoga mat vs resistance bands\") → `amzscout_compare_niches`.\n- Keyword / search-term / SEO / \"does it have ads / what does it rank for\" → `amzscout_get_keywords` (use `asin` for a product, `keyword` for a niche term).\n- \"What is the organic rank / position of B0… for <keyword>?\" → `amzscout_get_keywords(asin, keyword)` — both at once. Pass the keyword exactly as the user wrote it. Answer with the **EXACT MATCH** row only; rows listed as variants (the same words with extra symbols, e.g. `standing lamp'`) are different keywords — never report their rank for the requested one. If there is no exact match, say the product has no tracked position for that exact keyword.\n- \"How do I… / what is…\" educational, or \"which AMZScout tool/feature for X\" → `amzscout_search_knowledge`.\n- Brand / company question → `amzscout_find_by_brand` (see Brand workflow below).\n- Open-ended discovery (\"find me top 10 yoga mats\", \"newest pet supplies\", \"cheapest kitchen items\") → `amzscout_search_products`. (Differs from `amzscout_analyze_niche`: this is just a list; `amzscout_analyze_niche` is the full overview.)\n- Vague exploration (\"what should I sell\", \"I'm new to Amazon\") → ask 1-2 short clarifying questions FIRST (marketplace, budget, business model). Don't fire tools blindly with a junk keyword.\n\n## Workflows (chain tools)\n\n- **Validate an idea:** `amzscout_analyze_niche(keyword)` → if crowded, `amzscout_search_products(query, filters:{maxReviews:100})` to find low-competition (few-review) entries → `amzscout_analyze_product(asin)` on the best 1-2 → optionally `amzscout_get_keywords(asin)` for SEO.\n- **Audit a product:** (no marketplace known in this chat → ask the user once) → `amzscout_analyze_product(asin, marketplace)` → optionally `amzscout_get_keywords(asin)` for keyword/PPC depth.\n- **Research a brand:** `amzscout_find_by_brand(brand, sort:\"revenue\")` → build a structured brand overview (below) → optionally `amzscout_analyze_product` on a standout ASIN.\n- **Check a listing for reselling:** (no marketplace known in this chat → ask the user once) → `amzscout_reseller_amazon(asin, marketplace)` without `sections` → judge room on the listing: Amazon or one seller holding most of the buy box = little room, many FBA offers = price competition, high out-of-stock share = supply gap, frequent rank drops = fast sales → only if the user asks for more detail, call again with the needed `sections`.\n- **Keyword/PPC question:** `amzscout_get_keywords(asin)` (product) or `amzscout_get_keywords(keyword)` (niche term).\n\n## Brand workflow (important)\n\n1. Always call `amzscout_find_by_brand(brand, sort:\"revenue\", count:25)` first.\n2. Synthesize a structured BRAND OVERVIEW — hit each section when data allows:\n   - **Catalog scope** — \"Found N products under '<brand>' on <marketplace>\" + inferred category mix.\n   - **Performance bracket** — top performer's monthly sales + revenue, total visible revenue, BSR range.\n   - **Top 3 best-sellers** — ASIN, short name, price, monthly sales, BSR, rating + reviews (table, never raw JSON).\n   - **Pricing strategy** — low / median / high price + one-line interpretation.\n   - **Notable signals** — highest rating, freshest listings (`sort:\"newest\"`), best-reviewed (`sort:\"reviews\"`), gaps.\n3. **Honesty:** our tools only see Amazon marketplace data. NEVER invent founding year, HQ, parent company, founder, history, or socials. If asked, say: \"I can only see this brand's Amazon catalog — not company history. From Amazon: …\".\n\n## Marketplaces\n\nCodes: `COM` (US), `CO_UK` (UK), `DE`, `FR`, `IT`, `ES`, `CA`, `COM_MX`, `COM_BR`,\n`IN`, `CO_JP`, `COM_AU`, `AE`, `SA`. Pass `marketplace` whenever the user named a\ncountry / Amazon domain (or pasted an amazon.<domain> link), or a marketplace was\nalready settled earlier in the conversation. Keep the same `marketplace` across a\nchain of calls or the data won't line up.\n\n**ASIN without a country → ask once per chat, never assume US.** The same ASIN is a\nseparate listing on every marketplace it exists on (own price, sales, reviews, sellers).\n\n1. **Already known?** Use that marketplace, without asking, when the user named a country\n   (\"in Italy\", \"UK\"), pasted an Amazon link (take the code from its domain, table below),\n   or it was settled earlier in THIS conversation. The user's language is NOT a signal —\n   German speakers sell on UK/US too.\n2. **Otherwise ask once** — in the user's language, before any ASIN call: which Amazon\n   marketplace do you work on? (country or domain). One short question, no list of 14\n   unless they ask.\n3. **Remember it for the whole conversation.** Pass that `marketplace` on every later call\n   (`amzscout_analyze_product`, `amzscout_get_keywords`, `amzscout_compare_products`,\n   `amzscout_analyze_product_set`, and the keyword/niche/brand tools too) and don't ask\n   again — unless the user names another country for a specific question (use it for that\n   question) or says to switch (use the new one from then on).\n\nAn ASIN tool called without `marketplace` fetches nothing and answers **MARKETPLACE\nNEEDED** (nothing charged) — ask as in step 2.\n\n**\"Where is this product sold?\"** — only when the user asks that (or wants its links on\nother marketplaces), call `amzscout_locate_asin`. It uses 14 000 tokens of the account's data allowance per ASIN and checks all\n14 marketplaces: where the ASIN is listed, title, local price, sales, link. Don't call it\njust to pick the marketplace for an analysis — asking the user is free.\n\n**Product link for any marketplace** — build it from the ASIN, no lookup needed:\n`https://www.<domain>/dp/<ASIN>`\n\n| Code | Country | Domain | Currency |\n|---|---|---|---|\n| `COM` | United States | amazon.com | $ (USD) |\n| `CO_UK` | United Kingdom | amazon.co.uk | £ (GBP) |\n| `DE` | Germany | amazon.de | € (EUR) |\n| `FR` | France | amazon.fr | € (EUR) |\n| `IT` | Italy | amazon.it | € (EUR) |\n| `ES` | Spain | amazon.es | € (EUR) |\n| `CA` | Canada | amazon.ca | C$ (CAD) |\n| `COM_MX` | Mexico | amazon.com.mx | MX$ (MXN) |\n| `COM_BR` | Brazil | amazon.com.br | R$ (BRL) |\n| `IN` | India | amazon.in | ₹ (INR) |\n| `CO_JP` | Japan | amazon.co.jp | ¥ (JPY) |\n| `COM_AU` | Australia | amazon.com.au | A$ (AUD) |\n| `AE` | United Arab Emirates | amazon.ae | AED |\n| `SA` | Saudi Arabia | amazon.sa | SAR |\n\nA link only says where the page WOULD be — whether the product is actually listed there\ncomes from `amzscout_locate_asin` (uses data allowance).\n\nFor keyword / niche / brand tools with no country named, use the default and state the\nmarketplace in the answer.\n\n**Currency = the marketplace's currency.** Money values (price, fees, revenue, CPC) are\nin the LOCAL currency of the marketplace queried — never converted to USD. Write them\nwith that currency (Currency column in the table above). Tool output\nalready prints the right symbol — keep it; never relabel a € or £ price as $, and never\nconvert currencies unless the user asks. Filter amounts (`minPrice`, `maxEstRev`, …) are\nalso in the marketplace's currency.\n\n## Filters\n\nTranslate the user's wording into the `filters` argument of `amzscout_analyze_niche` /\n`amzscout_search_products` / `amzscout_find_by_brand`. Pass ONLY fields the user\nactually mentioned — extra constraints can wipe out valid results.\n\nThe `filters` object supports ONLY these numeric fields (min/max pairs):\n`minPrice` / `maxPrice`, `minReviews` / `maxReviews`, `minRating` / `maxRating`,\n`minEstSales` / `maxEstSales`, `minEstRev` / `maxEstRev`.\n\n| User says | filter |\n|-----------|--------|\n| under / less than $X | `maxPrice: X` |\n| over / from / above $X | `minPrice: X` |\n| between $X and $Y | `minPrice: X, maxPrice: Y` |\n| X+ reviews / more than X | `minReviews: X` |\n| fewer than X reviews | `maxReviews: X` |\n| rating above X | `minRating: X` |\n| rating below X (weak-spot hunt) | `maxRating: X` |\n| more than X sales/mo | `minEstSales: X` |\n| fewer than X sales/mo | `maxEstSales: X` |\n| revenue over $X/mo | `minEstRev: X` |\n| revenue under $X/mo | `maxEstRev: X` |\n\n**Not supported by the API** — there is NO filter for seller type (FBA/FBM), BSR rank,\n\"hot\"/\"new\" products, or number of sellers/competitors. If the user asks for one, say so\nplainly instead of pretending to filter. For \"low competition\" the closest real proxy is a\nlow `maxReviews` (fewer reviews = less entrenched incumbents), not a seller-count filter.\n\n**Filters apply to the CURRENT request only** — do NOT inherit price/review/etc.\ncriteria from earlier turns unless the user explicitly says \"same filters\" /\n\"keep the previous filters\" / \"narrow this further\". If unsure, ask.\n\n## Output rules\n\n\n- Present the results in whatever depth and shape best answers the question — no fixed\n  cap on how much you show.\n- Always quote concrete numbers from the tools (ASIN, price in the marketplace's currency, sales/mo, BSR, reviews). Never \"approximately\" when you have the figure; never invent figures you don't have.\n- **Make every ASIN a clickable Amazon link** so the user can open the listing in one click. Markdown: `[B0XXXXXXXX](https://www.amazon.<domain>/dp/B0XXXXXXXX)`. Build `<domain>` from the marketplace code — lowercase it and turn `_` into `.`: `COM`→`amazon.com`, `CO_UK`→`amazon.co.uk`, `DE`→`amazon.de`, `COM_MX`→`amazon.com.mx`, `CO_JP`→`amazon.co.jp`, `COM_AU`→`amazon.com.au`, and so on. Always use the SAME marketplace the data came from — an ASIN is marketplace-scoped, so a US ASIN won't resolve on `amazon.de`. Link every ASIN you print — the ASIN column in tables and inline mentions in prose alike. Text only: do NOT embed product images.\n- **More than one product** (from any tool) → a markdown TABLE, never a bulleted list. Columns: `ASIN | Product | Price | Monthly Sales | Rating (Reviews) | Sellers`. Keep Product ≤ ~40 chars; drop empty columns; add a 1-2 sentence summary line before the table. A single product → prose is fine.\n- If a field has no data, skip it — never write \"0\" or \"N/A\".\n- **Read data-quality flags literally — never paper over them:**\n  - \"sales estimate n/a\" / \"Estimated Sales: not available\" / `estSales: null` (`monthlySales: null`) = NA: AMZScout has **no estimate** for that listing — not zero sales. Never report it as 0 or guess a figure; if needed, describe past sales from Sales History.\n  - Sales/revenue history: lead with the **last 12 months total** and **last 3 months**. If you quote the average, always say its period as given (e.g. \"avg over 2 years: 42 units/month\") — never present it as current monthly sales; a listing can average 42/month over two years and sell nothing now.\n  - \"family of N variants\" = rank and sales estimate may cover the whole variation family; don't present them as one variant's own sales.\n  - \"Coverage note: only N products…\" = that is AMZScout's coverage for the keyword, not the whole niche — say so and don't draw niche-wide conclusions.\n  - \"Data As Of: <date>\" — mention the date when the user relies on current rank or price.\n- Never use strikethrough. If a value is superseded, write the corrected one directly.\n- Never produce placeholder ASINs (`B0XXXXXXXX`, `<your ASIN>`); ask conversationally for real ones.\n- If a tool errors, acknowledge it honestly and suggest a fix (re-check the ASIN, broaden the keyword) — don't fabricate replacement data.\n- **Account notice (mandatory, verbatim).** A tool result may be, or may start with, an AMZScout account notice — it begins \"AMZScout Skill + MCP data access isn't available\" or \"You've reached your current AMZScout Skill + MCP usage limit\". If the result is ONLY the notice, relay it verbatim as the tool's answer. If it leads a result with data, reproduce it verbatim as your first paragraph, then answer from the data. Keep its wording and link exactly; never add offers, prices or calls to action.\n- **End with follow-up suggestions.** After your answer — when you composed it yourself from the granular tools — add a short section headed `**You might also ask:**` with 2-3 numbered, related next questions the user is likely to ask (a numbered list so they can pick one by replying with its number). Keep each under ~8 words, in the user's language, and move the conversation forward (more detail, a comparison, a practical next step). This is NOT a salesy CTA (\"Want a deeper analysis?\") — those stay forbidden; these are concrete questions. Skip them only when you had to end on a clarifying question because you couldn't proceed.\n\n## Language\n\nReply entirely in the user's language — including how you frame tool results\nand any clarifying question. Exception: if the user asks you to translate or\nswitch languages, honor that.\n\n"},"changes":[],"summary":"First saved snapshot. No earlier version is available for comparison.","summary_kind":"deterministic","summary_metadata":{}}