Power your agent
The publisher's own URL. Not a PlugMCP endpoint.
Host a copy with our keys and logs.
https://mcp.canopyapi.co/mcpThe Claude button above opens Claude with the name and URL filled in. You still confirm it there.
claude mcp add --transport http co.canopyapi-mcp https://mcp.canopyapi.co/mcpRun in a terminal.
{ "mcpServers": { "co.canopyapi-mcp": { "command": "npx", "args": [ "-y", "mcp-remote", "https://mcp.canopyapi.co/mcp" ] } }}claude_desktop_config.json
Remote URL or a command on your machine? How the two installs differ
Run on PlugMCP
A new server you create. It is not this listing's official endpoint. Use the publisher install when you want their URL and their credentials.
LaunchFrom the publisher
An Amazon product data API served over the Model Context Protocol. Connect Claude, ChatGPT, Cursor, or any MCP client to real-time Amazon data — an Amazon search API, reviews API, price tracker, ASIN lookup, best sellers, deals, and sales estimates — powered by the Canopy API. No headless browsers, no brittle HTML scraping: typed JSON over standard MCP tools.
Amazon product API & MCP server: search, prices, ratings, sellers, deals, and ASIN lookup by Canopy.
Built with xmcp and deployed on Cloudflare Workers. Use the hosted remote MCP server at https://mcp.canopyapi.co/mcp or self-host from this repo. Setup guides and client walkthroughs live on the Amazon MCP server page.
AI agents and LLM apps increasingly need live e-commerce data. This server gives any MCP-capable assistant or agent framework direct tool access to Amazon:
Works as a Claude MCP server (Desktop, Code, and the claude.ai connectors directory), a ChatGPT MCP connector (Apps SDK, with rich product widgets), in Cursor and the MCP Inspector, and with any agent built on the MCP SDK, LangChain, or the OpenAI Agents SDK.
/mcp (current MCP spec)src/tools/openapi-typescriptAdd the remote MCP server to your client and sign in with OAuth, or supply a Canopy API key:
https://mcp.canopyapi.co/mcp
Claude and other OAuth-capable MCP clients will walk you through sign-in automatically. Get an API key at canopyapi.co, and see the Amazon MCP server guide for per-client setup instructions.
This repository is an Agent Plugin (specification 1.0.0). The plugin root is the repository root. A compatible client reads plugin.json, loads skills from skills/, and connects to the MCP server declared in mcp.json.
canopy-api-mcp/
├── plugin.json # Agent Plugins 1.0.0 manifest
├── mcp.json # hosted Streamable HTTP server
└── skills/
└── amazon-product-research/
├── SKILL.md
└── references/tools.md
mcp.json points at the same hosted endpoint as above:
{
"mcpServers": {
"canopy": {
"type": "streamable-http",
"url": "https://mcp.canopyapi.co/mcp"
}
}
}
Authorization stays in the client. OAuth-capable clients discover sign-in from the server. The plugin does not embed an API key: Agent Plugins treats headers as public package data, not a place for secrets. If a client only supports a static header, set CANOPY-API-KEY in that client.
Cursor. After the plugin is listed, install it from Customize or the Cursor Marketplace. To try the directory before it is listed, copy this repository to ~/.cursor/plugins/local/canopy-api-mcp (a symlink that points outside that folder is ignored), reload the window, and confirm the Canopy server and the amazon-product-research skill in Customize.
Other Agent Plugins clients. Load this directory. The client needs Agent Plugins 1.0.0 and the streamable-http transport. There is no archive and no registry protocol in the specification: the package is this directory.
server.json is unchanged. It is the MCP Registry server card, a separate format from plugin.json. npm install, npm run dev, and npm run deploy are unchanged.
Validate the package against the published schemas:
npm run validate:plugin
You'll need a Canopy API key from canopyapi.co.
npm install
npm run dev # xmcp watcher + wrangler dev (local Workers runtime)
npm run deploy # build + wrangler deploy --env production
Run npm test to build the worker and run the regression suite. Output-schema
tests use mocked REST responses to verify that null-valued fields (unavailable
prices, coupons, ratings) are stripped before reaching MCP clients, that
responses match the advertised JSON Schema, and that malformed values are still
rejected. These tests do not require a real Canopy API key.
Use the MCP Inspector and connect to http://localhost:8787/mcp (or your deployed URL). Provide the API key as a request header:
npx -y @modelcontextprotocol/inspector@latest
Or with curl:
curl -X POST http://localhost:8787/mcp \
-H "Content-Type: application/json" \
-H "Accept: application/json, text/event-stream" \
-H "CANOPY-API-KEY: $CANOPY_API_KEY" \
-d '{"jsonrpc":"2.0","id":1,"method":"tools/list","params":{}}'
Every request must carry credentials. Two modes are supported:
1. OAuth 2.1 (recommended for MCP clients). OAuth-capable clients discover the flow automatically: an unauthenticated request returns 401 with WWW-Authenticate: Bearer resource_metadata="https://mcp.canopyapi.co/.well-known/oauth-protected-resource". Supabase is the authorization server (dynamic client registration supported); consent lives at canopyapi.co. The verified user's Canopy API key is resolved server-side.
2. API key headers. Any of these forms works (checked in this order):
CANOPY-API-KEY: your-api-keyAPI-KEY: your-api-keyX-API-KEY: your-api-keyAuthorization: Bearer your-api-keyA missing or invalid credential returns HTTP 401 with a JSON-RPC error before the request reaches the MCP transport (see src/middleware.ts).
All 17 tools are read-only (readOnlyHint: true) and annotated with a title.
get_amazon_product — product details by ASIN, URL, or GTINget_amazon_product_variants — product variantsget_amazon_product_offers — seller offers and Buy Box infoget_amazon_product_stock — stock level estimatesget_amazon_product_sales — sales estimates (weekly, monthly, annual)get_amazon_product_top_reviews — top customer reviews (title, body, rating, helpful votes, images/videos)search_amazon_products — search with filters and sortingget_amazon_autocomplete — search term suggestionsget_amazon_deals — current dealsget_amazon_bestsellers — best-selling products for a categoryget_amazon_bestseller_categories — best seller category listget_amazon_categories — root category taxonomyget_amazon_category — category details with products and subcategoriesget_amazon_seller — seller information and product listingsget_amazon_author — author information and book listingsget_amazon_asin_from_gtin — ASIN lookup by ISBN/UPC/EANget_amazon_gtin_from_asin — GTIN lookup by ASINcanopy-api-mcp/
├── plugin.json # Agent Plugins 1.0.0 manifest
├── mcp.json # MCP server declaration (Streamable HTTP)
├── skills/
│ └── amazon-product-research/
│ ├── SKILL.md
│ └── references/tools.md
├── server.json # MCP Registry server card (separate from plugin.json)
├── LICENSE # MIT License
├── src/
│ ├── tools/ # one file per tool (auto-discovered by xmcp)
│ │ ├── get-amazon-product.ts
│ │ └── ...
│ ├── lib/
│ │ └── api-key.ts # reads API key from extra.authInfo
│ ├── middleware.ts # auth middleware (lifts header → authInfo.token)
│ ├── api-client.ts # type-safe Canopy REST client
│ └── types/
│ └── api.d.ts # generated from OpenAPI
├── xmcp.config.ts # xmcp config (endpoint, CORS, paths)
├── wrangler.jsonc # Cloudflare Workers config
├── worker-entry.ts # Workers entry (unchanged)
├── tsconfig.json
└── package.json
src/types/api.d.ts is generated from the Canopy OpenAPI spec. Regenerate when the API changes:
npm run generate
npm run dev — xmcp watcher + wrangler dev (local Workers runtime)npm run build — xmcp build --cf (emits worker.js for Cloudflare)npm run deploy — build + wrangler deploy --env productionnpm run delete — remove the deployed Workernpm run generate — regenerate API types from the OpenAPI specnpm run validate:plugin — check plugin.json, mcp.json, and skills/ against Agent Plugins 1.0.0v2.0 replaces ModelFetch with xmcp. The MCP endpoint moved from /sse/mcp (deprecated SSE transport mount) to /mcp (current Streamable HTTP convention). Update any clients accordingly.
This project is released under the MIT License. Copyright (c) 2026 Canopy API.
© 2026 MCP Directory.
Amazon product API & MCP server: search, prices, ratings, sellers, deals, and ASIN lookup by Canopy.
{ "mcpServers": { "co.canopyapi-mcp": { "url": "https://mcp.canopyapi.co/mcp", "headers": {} } }}Tools
No tool list yet. Remote servers get one after the first health check.
FAQ
Copy the http config for Cursor, Claude Desktop, Claude Code, or VS Code on this page. It uses the publisher's install, not a PlugMCP proxy.
No required environment variable is listed on this record.
Canopy API is reached over HTTP.
Last checked 2026-11-14. Amazon product API & MCP server: search, prices, ratings, sellers, deals, and ASIN lookup by Canopy.