5 Claude MCP Prompts for Amazon Product Research (Tested Live)
MCP (Model Context Protocol) is what lets Claude pull live data from an outside source instead of guessing. We covered the mechanics of that in our first article on AMZScout Skill+MCP — what it is, why ChatGPT and Claude can't see Amazon's marketplace on their own, and how the connection actually works.
This one skips the explainer and goes straight to Claude. We connected AMZScout Skill+MCP to a Claude account and ran five different types of research prompts against one real product niche — dog nail grinders — in a single conversation, so each answer builds on the one before it. Below is exactly what came back: full screenshots, plus what the numbers actually mean.
Quick Recap: What Claude MCP Actually Does
Model Context Protocol is an open standard, built by Anthropic, that lets Claude connect to an external tool over a live link instead of relying only on what it already knows. Once a connector — like AMZScout Skill+MCP — is added, Claude can call it mid-conversation, pull back current data, and answer with real numbers instead of an estimate. For the full breakdown of what that gap looks like without a connection, see our first article on AMZScout Skill+MCP.
Connecting AMZScout Skill+MCP in Claude
Setup is the same flow across claude.ai, Claude Desktop, and Cowork:
Buy a token pack and get your AMZScout Skill+MCP connector URL.
In Claude, open Settings > Connectors, click "Add custom connector," and paste the URL in.
Turn it on for a conversation from the Connectors panel next to the message box, then ask your question.
Connectors added through claude.ai or Claude Desktop also become available on mobile once enabled — there's nothing separate to install there.
One note before the examples: if you just want a quick check on a single product while browsing Amazon, the PRO AI Extension does that directly on the product page with no setup beyond installing it. Skill+MCP is for the research you're already doing through Claude — comparing, validating, and digging into a niche across a conversation, not a single page.
Five Prompts We Tested, One Real Niche
Everything below happened in one conversation on the same niche — dog nail grinders — so the results build on each other the way a real research session would: snapshot the market, audit a specific listing, compare it to the competitor actually winning, check its keywords, then ask Claude for a straight verdict. Screenshots are full and unedited.
1. Get the Market Snapshot
We asked Claude — connected via AMZScout Skill+MCP — for a market snapshot of the dog nail grinder niche: top products, revenue split, and who's actually dominant.
Claude's market snapshot for "dog nail grinder," pulled live via AMZScout Skill+MCP — 29 tracked listings, full brand breakdown.
Across 29 tracked listings, the niche does about $9.23M a month combined, roughly 324,000 units, at a median price of $26.99. The top 5 products hold 42% of that revenue — concentrated, but not winner-take-all, so there's room to compete without needing to unseat one single listing. LOPSIC is the clear leader: 7 listings pull almost half the niche's total revenue (49.98%) from under a quarter of the tracked ASINs. Casfuy has the most listings (8) but the lowest revenue per listing — more shelf presence than performance. Claude also caught three listings showing $0/mo in sales despite live pricing and flagged them as likely stockouts to exclude from any trend read, rather than quietly averaging them in — the same top-level view Product Database gives you inside Amazon, but through a conversation instead of a filter panel.
Revenue share by brand, charted from the same pull — LOPSIC's 7 listings out-earn Casfuy's 8.
2. Audit a Single Listing
Next we pointed Claude at one specific ASIN — Casfuy's flagship grinder, the most-reviewed listing in the niche — and asked for a straight read on whether it's a safe bet.
Full profile and sales history for ASIN B09KBX2VYJ, with Claude catching its own tool's mislabeled trend.
This is the useful part: AMZScout's own trend label called this listing "Declining Strongly" — but Claude caught why before repeating it. The label was being dragged down by the August 2026 data point, and the conversation happened only 10 days into that month, so that figure is a partial read, not a real drop. Once Claude excluded it, the actual pattern was a stable band of roughly 14,000–18,000 units and $275K–$330K a month from October 2025 through July 2026, with a seasonal spike in September 2025 and no sustained decline.
Its final verdict split the question in two. As a benchmark for content — citing the category leader in a roundup, say — yes, safe: 100,964 reviews, a #2 Claw Care rank, zero volatility. As a target to launch head-on against, no — that review count is a moat a new listing can't out-trust in any realistic timeframe, and Casfuy runs five near-identical variants at the same tier, all sitting on 100K+ reviews of their own. Claude pointed toward LOPSIC's cluster instead as the more interesting story — which is exactly what we asked it to compare next.
3. Put Two Listings Head-to-Head
Same niche, two listings, one direct question: which one is actually winning right now.
Casfuy vs. LOPSIC, compared live — price, sales, revenue, reviews, and 13 months of trend data.
Reviews alone would say Casfuy wins by a mile — 100,964 vs. 9,687. Every current performance metric says the opposite: LOPSIC sells 2.1x the units, pulls 2.6x the revenue, holds the #1 Claw Care rank against Casfuy's #2, and is the sole seller on its own listing with no buy-box competition. The trend history backs it up past the partial-month noise: Casfuy peaked in September 2025 and has held a flat $275K–$330K/month plateau since, while LOPSIC climbed from $288K to $746K over the same year — a 2.5x increase — while raising its price from $19 to $26. That's demand pulling the price up, not a discount-driven spike. Casfuy's edge is legacy trust built over five years; it just isn't converting into current sales momentum.
13 months of monthly revenue, plotted from Claude's own comparison data. The crossover is the whole story.
4. Pull the Keywords a Competitor Ranks For
Then we asked Claude to check LOPSIC's listing directly: what keywords is it ranking for, and is it leaving anything on the table?
Keyword coverage and gap analysis for ASIN B0C72Q1N5P, via AMZScout Skill+MCP — the same job Reverse ASIN Lookup does directly on AMZScout.
The listing has its core terms locked down — "dog nail grinder" (18,100 searches/month), "nail grinder for dogs" (14,800), "dog nail clippers" (12,100) — plus coverage most sellers wouldn't think to check on their own: it ranks for competitor-brand searches like "dremel dog nail grinder" and "vancoo pet nail trimmer," actively pulling traffic off rival brand names, and picks up real cross-species demand from "cat nail clippers" (74,000/month).
The gap Claude flagged is more specific than a missing broad term: "dog nail grinder (rechargeable, quiet)" pulls 6,880 searches a month and doesn't appear anywhere across the listing's 61 ranked keywords, even though the product is rechargeable. It checked the LED-light angle too and ruled it out — the "4 LED Lights" feature only maps to keywords with effectively zero search volume, so it's a nice differentiator on the page but not something to lead an SEO push with. It also flagged a real limitation unprompted: this pull returned search volume and CPC per term, not organic rank position, so it shows what the listing covers, not how strongly it ranks for each one.
Six of the 61 ranked keywords, plus the one 6,880-search term that isn't ranked at all.
5. Just Ask It in Plain English
Last, the question a seller would actually ask before committing to sourcing — no ASIN, no tool name, just the decision.
Claude's full entry-viability verdict for the dog nail grinder niche, synthesized from everything pulled earlier in the same conversation.
This is where the multi-turn part matters: Claude didn't just answer the saturation question, it pulled together everything from the four questions before it in the same conversation. It repeated the market size ($9.23M/month, 42% top-5 concentration) as the case for entering, then pointed to LOPSIC's trajectory — 9,600 reviews outselling a 100,964-review incumbent more than 2:1 — as proof review count isn't an insurmountable moat here if the product and positioning are right.
Then it drew the actual line: the generic "2-speed whisper-quiet grinder" lane is already defended from two directions, with Casfuy running 5+ near-identical variants and LOPSIC running roughly 7 of its own to own the SERP. The wedge it landed on was the same "rechargeable" keyword gap it had surfaced two questions earlier — 6,880 searches/month, unclaimed by any top listing — alongside size-specific and bundle angles. It also flagged its own blind spot without being asked: no visibility into landed cost, tariffs, or MOQ, which is what actually decides whether the margin math works. That's the kind of caveat a plausible-sounding guess wouldn't bother making.
FAQ
What is Claude MCP?
MCP (Model Context Protocol) is Anthropic's open standard for connecting Claude to outside tools and live data through a connector. Once a connector like AMZScout Skill+MCP is added, Claude can call it mid-conversation and answer with current data instead of an estimate.
How do I add an MCP server to Claude?
Open Settings > Connectors in Claude, click "Add custom connector," paste the server's URL, and authenticate. The flow is the same across claude.ai, Claude Desktop, and Cowork; connectors added on web or desktop also work on mobile once enabled there.
Does this work in Claude Desktop, or only claude.ai?
Both. AMZScout Skill+MCP is a remote MCP server, so it connects to Claude from Anthropic's infrastructure rather than your device — it works the same in claude.ai, Claude Desktop, or Cowork.
Do I need to be technical to set this up?
No. Adding a connector means pasting a URL into a settings panel — no code, no config file to edit.
Pricing
AMZScout Skill+MCP runs on token packs rather than a subscription, active for 30 days from purchase — plans start at $39 for 1M tokens (about 50 product analyses). Full pricing and pack details are in our first article on AMZScout Skill+MCP.
The Bottom Line
Every example above ran in one real conversation, on one real niche, through AMZScout Skill+MCP: a market snapshot, a listing audit, a head-to-head, a keyword pull, and one plain-English strategic question that pulled all four together — same connector, five different ways to ask it something useful.
Connect it once, and the next research question you ask Claude gets answered with AMZScout's actual data behind it, not a guess dressed up as an answer.
