npm install lumnix

🎬 Demos

Why Lumnix

The old way: Subscribe to Jungle Scout for product research ($49/mo), Keepa for price history ($50/mo), some Chrome extension for Alibaba vetting ($20/mo), and a spreadsheet for margin math. Switch between tabs all day. Miss opportunities.

The Lumnix way: One API key. 19 tools. In your AI assistant. No dashboards. No Chrome extensions. No tab-hopping.

"Find me wireless earbuds under $20 with high sales volume, low reviews, and a supplier on Alibaba with MOQ under 500"

Your AI calls Lumnix. You get structured data with revenue estimates, competition analysis, and profit margins. You decide. Done.

What makes it different

• 🚫 No dashboards — Works inside Claude, Cursor, Windsurf, Cline, or any MCP client
• 📊 Keepa-powered — Real price history and BSR tracking, not estimates
• 🎯 Opportunity Score — 0-100 niche score based on demand, competition, and margin potential
• 🏭 Full sourcing pipeline — Amazon → Alibaba → AliExpress, all connected
• 💰 Profit calculator — Full P&L with FBA fees, shipping, PPC ads, refunds, and ROI
• ⚡ 30-second setup — No install. No config file editing. Just an API key and one JSON block

⚡ Quick Start

1. Get an API key (free)

Sign up at lumnix.dev — no credit card required. Get your key in 30 seconds.

2. Add to your MCP client

Claude Desktop — edit ~/Library/Application Support/Claude/claude_desktop_config.json:

{
  "mcpServers": {
    "lumnix": {
      "command": "npx",
      "args": [
        "mcp-remote",
        "https://lumnix.dev/api/mcp",
        "--header",
        "Authorization: Bearer YOUR_API_KEY"
      ]
    }
  }
}

⚠️ Claude Desktop doesn't support remote MCP URLs natively. The mcp-remote bridge (auto-installed by npx) makes it work. Requires Node.js installed.

Cursor — Settings → MCP → Add new global MCP server:

{
  "mcpServers": {
    "lumnix": {
      "url": "https://lumnix.dev/api/mcp",
      "headers": {
        "Authorization": "Bearer YOUR_API_KEY"
      }
    }
  }
}

Windsurf / Cline / any MCP client — same pattern as Cursor. URL + Bearer token header.

3. Start researching

Restart your AI client. Then just talk naturally:

"Search Amazon for portable blenders with high sales and low reviews"

That's it. No Docker. No config files to debug.

🛠 Tools Reference

Lumnix gives your AI 19 specialized tools across three platforms:

Amazon — 11 tools

Alibaba — 4 tools

AliExpress — 4 tools

💬 Example Prompts

Product Research

"Search Amazon for stainless steel water bottles between $15-30 with at least 4 stars and under 1000 reviews"

Opportunity Scoring

"Score the 'portable blender' niche on Amazon. Is it worth entering?"

Competition Analysis

"Analyze the competition for 'organic dog treats'. How many sellers have less than 500 reviews?"

Keepa Price History

"Show me the price history for ASIN B0FHWKYT89. Has it been trending down?"

Supplier Vetting

"Find aluminum water bottle suppliers on Alibaba with trust score above 70 and MOQ under 500"

Margin Calculation

"I want to sell this AliExpress product on Amazon for $29.99. Calculate my profit margin including FBA fees, PPC ads at 15%, and 5% refund rate"

Full Workflow

"Find a product on Amazon with high sales volume and low reviews, find the matching supplier on Alibaba, vet them, and tell me if the margins work"

📊 Example Response

{
  "asin": "B0FHWKYT89",
  "title": "Portable Blender for Shakes and Smoothies",
  "price": "$29.99",
  "rating": 4.5,
  "reviews": 2847,
  "sales_volume": "5K+ bought in past month",
  "monthly_revenue_est": "$149,950",
  "is_amazon_choice": true,
  "is_prime": true,
  "number_of_offers": 12
}

💰 Plans

The free tier gives you 3 Amazon search tools — enough to see the data quality. When you're ready for Keepa-powered insights, opportunity scoring, and full sourcing across all platforms, upgrade here.

🔧 API Reference

Lumnix is a remote MCP server. No installation, no Docker, no local dependencies.

Endpoint: https://lumnix.dev/api/mcp Auth: Bearer token in the Authorization header Protocol: JSON-RPC 2.0 (MCP spec 2025-03-26)

Direct API call

curl -X POST https://lumnix.dev/api/mcp \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer YOUR_API_KEY" \
  -d '{
    "jsonrpc": "2.0",
    "id": 1,
    "method": "tools/call",
    "params": {
      "name": "amazon_search_products",
      "arguments": { "keyword": "wireless earbuds", "max_price": 50 }
    }
  }'

🌐 Find Us On

Lumnix is listed on all major MCP directories:

• Smithery.ai — Search "lumnix"
• MCP Servers — Search "lumnix"
• Claude MCP Directory — Coming soon
• npm — npm install lumnix · npm info lumnix

❓ FAQ

What MCP clients work with Lumnix? Any client that supports remote MCP servers with HTTP transport: Claude Desktop, Cursor, Windsurf, Cline, VS Code (with MCP extension), and more.

Is the free tier really free? Yes. 10 API calls/month, no credit card. Enough to search products and see the data quality before committing.

Where does the data come from? Amazon data via web APIs + Keepa (for price/BSR history). Alibaba and AliExpress via their respective platforms. All structured, enriched, and ready to act on.

How fast are the responses? Most tools respond in 1-5 seconds. Deep analysis tools (competition, reviews, opportunity score) can take up to 15 seconds.

Can I use Lumnix in my own app? Yes. The REST API works with any HTTP client. Call the MCP endpoint directly with your API key — no AI assistant required.

Do I need to install anything? No. Lumnix is a remote server. Just add the config to your MCP client and go.

🔒 Security

• API keys hashed with SHA-256 — never stored in plaintext
• Row-level security on all database tables
• Per-platform rate limiting
• Webhook signature verification for all payment events

📄 License

MIT