Shop2LLM API & MCP Docs

MCP Server

The Model Context Protocol (MCP) endpoint lets AI assistants (ChatGPT, Claude, Gemini, and any MCP client) search real product data across connected stores. Requests are JSON-RPC 2.0 over HTTP.

Tools

• search_products — search products across all connected stores. Parameters: query (string, required), limit (integer, optional — default 10, max 50).
• get_product — fetch full details for one product. Parameters: product_id (string, required) and store_id (string, required — both are returned by search_products).

Example request

curl -X POST https://api.shop2llm.com/mcp \
  -H "Content-Type: application/json" \
  -H "Accept: application/json, text/event-stream" \
  -d '{
    "jsonrpc": "2.0",
    "id": 1,
    "method": "tools/call",
    "params": {
      "name": "search_products",
      "arguments": { "query": "running shoes", "limit": 10 }
    }
  }'

MCP clients authenticate with OAuth 2.1 (authorization code flow with PKCE) or an API key — both are handled by the gateway.

Product Feed

Every license key exposes a Google-compatible CSV product feed. The license key in the path acts as the credential — no extra auth headers needed.

• CSV format (product_id, title, description, link, image_link, availability, price, brand, condition, id), up to 500 products, served with a 1-hour cache (Cache-Control: max-age=3600).

JSON statistics for the same feed — store URL, product count, currency, and data-quality counts (missing image / price / description).

curl https://api.shop2llm.com/fincore/v1/feed/{LICENSE_KEY}/info

AI Visibility Check

Run the public AI visibility analyzer against any store URL. No account required.

curl -X POST https://api.shop2llm.com/fincore/v1/ai-check/analyze \
  -H "Content-Type: application/json" \
  -d '{"url":"https://store.com"}'

The response includes an overall score (0–100), a level (excellent / good / fair / poor), and a factors summary — llms.txt presence, AI-crawler rules in robots.txt, JSON-LD schema types, meta description, sitemap, and per-engine detection.

Rate limit: 20 requests per hour per IP. Exceeding it returns HTTP 429 with {"error":"rate_limited"}.

Rate limits & auth

Session-authenticated endpoints accept the token either as a standard Authorization: Bearer <token> header or as X-Shop2LLM-Token: <token> — both are equivalent; the latter is useful where proxies strip the Authorization header.

A 429 response means the windowed counter for your IP bucket is full — retry after the window resets. Nothing is charged and no quota is consumed.