Shopping agents are one of the most requested capabilities in AI applications. Users want assistants that can find real products, compare prices across markets, and recommend what to buy — not just generate placeholder text.
But building a shopping agent from scratch means scraping storefronts, normalizing product data, handling region-specific availability, and maintaining fragile parsers. That is weeks of work before you even get to the agent logic.
Enter MCP (Model Context Protocol) and the BuyWhere MCP server. In 10 minutes you can give any MCP-compatible client real cross-border product search, price comparison, and merchant discovery.
What you will build
An AI agent that can:
- Search real products across Singapore, SEA, and US markets
- Compare prices across merchants in a single query
- Answer "where can I buy this?" with actual merchant links
All without scraping a single storefront.
Step 1: Get your API key
Sign up at buywhere.ai/api-keys — no sales call, no credit card. You get a live API key instantly.
Step 2: Add the MCP server to your client
The BuyWhere MCP server is published as @buywhere/mcp-server on npm. Zero installation required — just configure your MCP client.
Claude Desktop
Add to claude_desktop_config.json:
{
"mcpServers": {
"buywhere": {
"command": "npx",
"args": ["-y", "@buywhere/mcp-server"],
"env": { "BUYWHERE_API_KEY": "bw_live_your_key_here" }
}
}
}
Cursor / VS Code / Cline / OpenCode
Same configuration — add to your MCP settings file:
{
"mcpServers": {
"buywhere": {
"command": "npx",
"args": ["-y", "@buywhere/mcp-server"],
"env": { "BUYWHERE_API_KEY": "bw_live_your_key_here" }
}
}
}
Step 3: Start querying
Once configured, your MCP client automatically exposes the BuyWhere tools. Here is what you can ask:
Search products across markets
"Find wireless noise-cancelling headphones available in Singapore under $200"
The agent calls search_products with your criteria and returns real product listings with prices, merchants, and availability.
Compare prices
"Compare the top 3 headphones — which is the best value?"
The agent calls compare_prices and returns a structured side-by-side comparison.
Multi-region search
"What are the best options for this product in Singapore vs the US?"
BuyWhere supports explicit region filtering, so your agent can answer market-specific shopping queries.
Available tools at a glance
| Tool | What it does |
|---|---|
search_products |
Search by keyword, category, price range, region |
get_product |
Full details by product ID (prices, specs, images) |
compare_prices |
Side-by-side comparison of 2–5 products |
get_price |
Current prices across all merchants for one product |
get_affiliate_link |
Click-tracked affiliate URL |
get_catalog |
Available product category taxonomy |
Why MCP instead of a custom tool integration?
MCP standardizes how agents discover and call tools. One server works across Claude Desktop, Cursor, VS Code Copilot, Cline, OpenCode, and any MCP-compatible client. You write the configuration once, and your agent has shopping superpowers everywhere.
Why BuyWhere instead of scraping or platform APIs?
- Scraping breaks when sites change and requires per-storefront logic
- Amazon APIs only return Amazon products
- Google Shopping is not a normalized product layer you control
BuyWhere is a neutral, cross-merchant product catalog. We index products from merchants across Singapore, SEA, and US markets into one structured API. No inventory to sell, no platform to favour.
Next steps
- Get your API key (60 seconds)
- Add the MCP config (2 minutes)
- Start building your shopping agent (7 minutes)
Check the full documentation and the GitHub repo for advanced usage, batch queries, and production deployment patterns.
BuyWhere is in developer beta. Coverage: Singapore, Southeast Asia, and growing US market. MCP server: @buywhere/mcp-server
Related reading:
Top comments (0)