Amazon Product Intelligence MCP Server

An Apify Actor that runs as a persistent Model Context Protocol (MCP) server, turning Amazon into actionable seller intelligence for AI assistants.

What Makes This Unique

Existing Amazon scrapers return raw data dumps. This server returns intelligence — every product and keyword response includes metrics that answer the question "Should I sell this?"

1. Opportunity Score (0–100)

A composite score calculated from five signals:

Component	Weight	Signal
Demand	25%	Review count (proxy for cumulative sales)
Competition	30%	Number of sellers + Amazon selling penalty
Price Health	20%	Margin room by price tier
Review Velocity	15%	Growth stage proxy
BSR	10%	Best Sellers Rank quality

Formula:

score = demand×0.25 + competition×0.30 + priceHealth×0.20 + velocity×0.15 + bsr×0.10

Clamped to [0, 100]. Labels: excellent (80+), good (60–79), moderate (40–59), weak (20–39), avoid (<20).

2. Niche Gap Detection (search_products)

Flags when ≥60% of page-1 results have fewer than 100 reviews — a reliable signal that competition is thin and a new seller can rank organically without massive ad spend.

3. FBA Profitability Estimate (analyze_product)

Rough but fast margin math using Amazon's 2024 US FBA fee schedule:

• Referral fee — category-specific rate (8–20%)
• Fulfillment fee — size-tier based ($3.06–$89.98+)
• Storage fee — per-unit monthly estimate ($0.87/ft³, Jan–Sep)
• CoGS — configurable % of sale price (default 25%)
• Net margin — USD and % after all fees

---

MCP Tools

search_products

Search Amazon for products matching a keyword.
Returns up to max_results products from page 1, each with
Opportunity Score and niche gap detection for the keyword.

Parameters:

Name	Type	Required	Description
keyword	string	✅	Amazon search keyword (e.g. "yoga mat")
category	string	❌	Category hint for FBA fee lookups
max_results	integer 1–20	❌	Number of products (default: 10)

Response fields include:

• nicheGap.detected — boolean niche gap flag
• nicheGap.signal — human-readable explanation
• Per product: opportunityScore.total, opportunityScore.breakdown, full product data

---

analyze_product

Deep-analyse a specific Amazon product by ASIN.
Returns full product data, Opportunity Score breakdown, and FBA profitability estimate.

Parameters:

Name	Type	Required	Description
asin	string	✅	10-char Amazon ASIN (e.g. "B07PXGQC1Q")
target_price	number	❌	Override price for FBA calculation
estimated_cogs_pct	number 0–1	❌	CoGS fraction (default: 0.25)
category	string	❌	Category override for referral fee

Response includes:

• Full product data (title, price, rating, review count, BSR, seller info)
• opportunityScore with full breakdown
• fbaEstimate with itemised fees and net margin

---

Connecting to Claude (or any MCP client)

This Actor runs as a long-lived web server (standby mode). After starting the Actor:

1. Note the Actor's web server URL from the Apify console
2. Add it to your MCP client configuration:

{
  "mcpServers": {
    "amazon-intel": {
      "url": "https://<your-actor-run-url>/sse"
    }
  }
}

Or use Claude Desktop's MCP settings to point to the SSE URL.

Endpoints

Endpoint	Method	Description
/sse	GET	MCP SSE connection endpoint
/messages?sessionId=<id>	POST	MCP message endpoint
/health	GET	Health check

---

Configuration

Proxies (Recommended)

Amazon aggressively blocks scrapers. For production use, enable Apify Residential Proxies in the Actor input:

{
  "proxyConfiguration": {
    "useApifyProxy": true,
    "apifyProxyGroups": ["RESIDENTIAL"]
  }
}

Without residential proxies you may encounter CAPTCHAs, especially at high request volumes.

---

Example Prompts

Once connected to Claude or another MCP AI:

"Find me low-competition products in the kitchen niche with good margins"

"Analyze ASIN B07XJ8C8F5 — is it worth entering this niche?"

"Search for 'bamboo toothbrush' and tell me if there's a niche gap"

"I want to sell a product for $35. Analyze B09Q3HZTHZ and model the FBA fees assuming 20% COGS"

---

Limitations & Disclaimers

• Scores are estimates, not guarantees. Validate with Amazon Seller Central before sourcing.
• FBA fees change — verify current rates at sellercentral.amazon.com.
• Seller count is estimated from public buybox data; actual FBA seller count may differ.
• CoGS estimate (default 25%) is a rough placeholder; negotiate your own supplier pricing.
• This tool scrapes publicly available data. Use in compliance with Amazon's terms of service.

---

Technical Details

• Runtime: Node.js 20, TypeScript (ESM)
• Scraping: Playwright + Chromium (headless)
• Protocol: MCP over SSE (@modelcontextprotocol/sdk)
• Proxy: Apify residential proxy rotation
• Docker base: apify/actor-node-playwright-chrome:20