The Amazon MCP server for AI agents
Give Claude, Cursor, or any MCP client live Amazon product data. Product lookup, search, pricing, deals, and best sellers, hosted and ready with one URL.
One URL, no install
The server is hosted at mcp.canopyapi.co and speaks Streamable HTTP. Add it from the command line in Claude Code, or drop the JSON config into Claude Desktop, Cursor, or any other MCP client. Sign in with OAuth when prompted, or use your API key as a bearer token.
claude mcp add --transport http canopy https://mcp.canopyapi.co/mcp
{
"mcpServers": {
"canopy": {
"url": "https://mcp.canopyapi.co/mcp"
}
}
}What is an Amazon MCP server?
MCP, the Model Context Protocol, is the open standard AI assistants use to call external tools. An Amazon MCP server turns Amazon data into tools an agent can call on its own: look up a product by ASIN, search the catalog, check current offers, list the best sellers in a category.
Without one, an AI shopping agent is guessing from training data that may be years stale. With one, every answer about price, availability, or ranking comes from a live request. That is the difference between a chatbot that talks about agentic commerce and an agent that actually does it.
The Canopy Amazon MCP server is backed by the same infrastructure as our Amazon product API: live data for any of 350 million products, no scraping, no proxies, and no Amazon approval process.
What your agent can pull
- Product lookup
- Full product details by ASIN or GTIN/UPC: title, brand, images, variants, descriptions, and A+ content.
- Search and rankings
- Keyword search results with positions, so agents can compare products or track where a listing ranks.
- Pricing and offers
- Real-time pricing, Buy Box data, and seller offers, plus sales and stock estimates for demand signals.
- Deals and best sellers
- Current deals, new releases, and best-seller lists across more than 25,000 categories.
Hosted MCP server vs. building your own
You could wire an agent to Amazon data yourself with a scraper or SP-API integration wrapped in a local MCP server. Here is what that trade looks like.
| Canopy Amazon MCP | Build it yourself | |
|---|---|---|
| Setup | Add one URL to your MCP client | Write and host a server, manage scraping or SP-API credentials |
| Amazon approval | None needed, just an API key | SP-API requires a seller account and approval; scraping requires proxy management |
| Data coverage | Product details, search, offers, sales and stock estimates, categories, sellers, deals, best sellers | Whatever you build endpoint by endpoint |
| Maintenance | Hosted, versioned, and monitored for you | Yours to patch every time Amazon changes markup or policies |
| Safety | All tools are read-only | Depends on what you expose |
| Other interfaces | Same key works for REST, GraphQL, and LLM skills | MCP only, unless you build more |
Your MCP client discovers the current tool list automatically after connecting. Machine-readable details live in the server card at canopyapi.co/.well-known/mcp/server-card.json.
Built for AI shopping agents and agentic commerce
Agentic commerce is moving from demo to production: assistants that research products, compare offers, and monitor prices on a schedule. Those agents are only as good as the data they can reach at run time.
Teams use the Canopy Amazon MCP server to power shopping copilots that shortlist products from a natural-language brief, repricing and market-research agents that watch competitors, and internal assistants that answer catalog questions straight from live Amazon data. The same tools also work in Claude Skills and scheduled agent workflows, so a one-off question and a recurring monitor use the same integration.
Frequently asked questions
What is an Amazon MCP server?
MCP (Model Context Protocol) is an open standard that lets AI assistants call external tools. An Amazon MCP server exposes Amazon data as tools an agent can call directly, so instead of copying product pages into a chat, Claude or another assistant can look up products, run searches, and pull pricing on its own. Canopy hosts one at mcp.canopyapi.co/mcp, backed by the same data as its REST and GraphQL APIs.
Does the Amazon MCP server work with Claude?
Yes. It works with Claude Desktop, Claude Code, and claude.ai connectors, plus Cursor, and any other client that supports remote MCP servers over Streamable HTTP. You add one URL and the client discovers the tools automatically.
Is this the official Amazon MCP server?
No. Amazon publishes its own MCP servers for other jobs, such as AWS infrastructure, Amazon Q, and Amazon Ads. Canopy API is an independent service focused on one thing: live Amazon retail catalog data, including product details, search, pricing, deals, and best sellers.
How do I authenticate?
Two options: sign in with OAuth 2.1 directly from your MCP client using your Canopy account, or send your API key as a bearer token. Every account includes 100 free requests per month, and one key works across MCP, REST, and GraphQL.
Can I get Seller Central or SP-API data through this MCP server?
No. Seller Central account data, orders, and inventory require Amazon SP-API access tied to your own seller account. The Canopy Amazon MCP server covers the public retail catalog: any of 350 million products, search results, offers, categories, deals, and best sellers, with no Amazon approval process required.
What can an AI shopping agent do with it?
Typical agentic commerce flows include finding products that match a natural-language brief, comparing prices and Buy Box offers across candidates, monitoring deals in a category, and enriching a product catalog with live Amazon data. All tools are read-only, so the agent can research freely without side effects.
What does it cost?
The same pricing as the rest of Canopy API: 100 free requests every month, then $0.01 per request pay-as-you-go with automatic volume discounts. MCP tool calls are billed like any other API request.
Plug your agent into Amazon
Create an account, connect the MCP server with one URL, and your agent is pulling live Amazon product data in minutes. 100 requests free every month.