merchGent

[](https://www.npmjs.com/package/merch-connector) [](LICENSE) [](https://nodejs.org) [](https://modelcontextprotocol.io)

An MCP server that gives AI agents eyes on any e-commerce storefront.

Scrape product listings, extract facets, badges, sort options, and B2B signals; run AI-powered merchandising audits; compare two storefronts side-by-side; detect what changed between visits; and build persistent memory about sites — all through the Model Context Protocol.

---

E-commerce merchandising analysis is manual, repetitive, and fragmented. A merchandiser might spend hours clicking through competitor sites, checking if filters work, comparing product grids, and noting what's changed. AI agents can do this work — but they can't see storefronts the way shoppers do.

merch-connector bridges that gap. It gives any MCP-compatible AI agent (Claude, custom agents, etc.) the ability to:

• Browse any storefront with a stealth headless browser that handles bot protection
• Extract structured product data, facets, performance metrics, and page structure
• Analyze merchandising quality through five expert personas or a full roundtable debate
• Remember site quirks across sessions so the agent gets smarter over time
• Track changes across visits — new products, price moves, facet/sort changes

---

npx merch-connector

The server communicates over stdio and is designed to be launched by an MCP client, not run standalone.

Add to your Claude Desktop claude_desktop_config.json or Claude Code .mcp.json:

{
  "mcpServers": {
    "merch-connector": {
      "command": "npx"