mcp-amazon
Enables AI agents to search products, manage cart, place orders, and track shipments on Amazon via browser automation.
README
@striderlabs/mcp-amazon
Amazon MCP Server by Strider Labs — give AI agents the ability to search products, manage your cart, place orders, and track shipments on Amazon via browser automation.
Features
- Product Search — search Amazon with full result details (ASIN, price, ratings, Prime badge)
- Product Details — fetch complete product info by ASIN or URL
- Cart Management — add items, view cart, clear cart
- Order Placement — preview and place orders with dual-confirmation safety
- Order Tracking — track shipments with status and carrier info
- Order History — retrieve past orders
- Prime Status — check Prime membership and benefits
- Address Management — set delivery address
- Session Persistence — cookies saved to
~/.strider/amazon/across sessions - Stealth Mode — random delays, user agent rotation, webdriver masking to reduce bot detection
Installation
npm install -g @striderlabs/mcp-amazon
# Install Playwright browsers (required)
npx playwright install chromium
MCP Client Configuration
Claude Desktop (~/.claude/claude_desktop_config.json)
{
"mcpServers": {
"amazon": {
"command": "striderlabs-mcp-amazon"
}
}
}
Generic MCP Client
{
"mcpServers": {
"amazon": {
"command": "npx",
"args": ["-y", "@striderlabs/mcp-amazon"]
}
}
}
Available Tools
| Tool | Description |
|---|---|
amazon_status |
Check login status and session info |
amazon_login |
Initiate login flow (returns URL + instructions) |
amazon_logout |
Clear session and cookies |
amazon_search |
Search products by keyword |
amazon_get_product |
Get product details by ASIN or URL |
amazon_add_to_cart |
Add product to cart (ASIN, URL, or query) |
amazon_view_cart |
View cart contents and subtotal |
amazon_clear_cart |
Remove all items from cart |
amazon_preview_order |
Preview order before placing |
amazon_place_order |
Place order (requires confirm=true) |
amazon_track_order |
Track order by ID |
amazon_get_orders |
Get order history |
amazon_prime_check |
Check Prime membership status |
amazon_set_address |
Set delivery address |
Usage Examples
Search and add to cart
Search for "AirPods Pro" on Amazon
→ amazon_search({ query: "AirPods Pro", maxResults: 5 })
Add the first result to cart
→ amazon_add_to_cart({ asinOrQuery: "B0BDHWDR12", quantity: 1 })
Place an order
Preview my order before placing
→ amazon_preview_order()
Place the order (user must confirm first)
→ amazon_place_order({ confirm: true })
Track a shipment
Track order 123-4567890-1234567
→ amazon_track_order({ orderId: "123-4567890-1234567" })
Authentication Flow
Amazon requires manual login (headless browsers cannot complete Amazon's full auth flow including 2FA):
- Run
amazon_login— it returnshttps://www.amazon.com/ap/signin - Open that URL in your browser and log in with your Amazon credentials
- Complete any 2-factor authentication
- Run
amazon_statusto verify the session is detected - Cookies are saved to
~/.strider/amazon/cookies.jsonfor future sessions
Configuration
Session data is stored in:
~/.strider/amazon/
├── cookies.json # Browser session cookies
└── session.json # Cached session metadata
To reset: run amazon_logout or delete the ~/.strider/amazon/ directory.
Order Safety
amazon_place_order requires confirm: true to actually place an order:
- Without
confirm: true— returns a preview (same asamazon_preview_order) - With
confirm: true— places the order immediately
Never set confirm: true without explicit user consent.
Bot Detection
Amazon has aggressive bot detection. This server includes:
- Random delays between actions (500–2000ms)
- User agent rotation (Chrome/Safari on macOS/Windows)
- Webdriver property masking
- Browser plugin/language spoofing
- Proper viewport and screen resolution
If you receive a CAPTCHA error: visit amazon.com in your regular browser, complete the CAPTCHA, then retry. Cookies will be re-used in subsequent calls.
Technical Details
- Runtime: Node.js 18+, ES modules
- Browser: Playwright Chromium (headless)
- Transport: MCP stdio
- Session storage:
~/.strider/amazon/
Limitations
- Requires manual login (no programmatic auth for Amazon)
- CAPTCHA challenges must be completed in a real browser
- Payment method must be pre-configured on your Amazon account
- Amazon's UI changes may break DOM selectors — report issues on GitHub
- Not affiliated with Amazon.com, Inc.
Troubleshooting
"CAPTCHA detected" Visit amazon.com in your browser, complete the CAPTCHA, then retry.
"Not logged in"
Run amazon_login, open the URL, log in, then run amazon_status.
"Add to Cart button not found"
The item may be sold by a third-party seller requiring seller selection, or it may be unavailable. Try amazon_get_product first to verify availability.
"Page load timed out" Check your internet connection. Amazon pages can be slow on first load.
Development
git clone https://github.com/striderlabs/mcp-amazon
cd mcp-amazon
npm install
npx playwright install chromium
npm run build
npm start
License
MIT — see LICENSE
Built by Strider Labs
推荐服务器
Baidu Map
百度地图核心API现已全面兼容MCP协议,是国内首家兼容MCP协议的地图服务商。
Playwright MCP Server
一个模型上下文协议服务器,它使大型语言模型能够通过结构化的可访问性快照与网页进行交互,而无需视觉模型或屏幕截图。
Magic Component Platform (MCP)
一个由人工智能驱动的工具,可以从自然语言描述生成现代化的用户界面组件,并与流行的集成开发环境(IDE)集成,从而简化用户界面开发流程。
Audiense Insights MCP Server
通过模型上下文协议启用与 Audiense Insights 账户的交互,从而促进营销洞察和受众数据的提取和分析,包括人口统计信息、行为和影响者互动。
VeyraX
一个单一的 MCP 工具,连接你所有喜爱的工具:Gmail、日历以及其他 40 多个工具。
graphlit-mcp-server
模型上下文协议 (MCP) 服务器实现了 MCP 客户端与 Graphlit 服务之间的集成。 除了网络爬取之外,还可以将任何内容(从 Slack 到 Gmail 再到播客订阅源)导入到 Graphlit 项目中,然后从 MCP 客户端检索相关内容。
Kagi MCP Server
一个 MCP 服务器,集成了 Kagi 搜索功能和 Claude AI,使 Claude 能够在回答需要最新信息的问题时执行实时网络搜索。
e2b-mcp-server
使用 MCP 通过 e2b 运行代码。
Neon MCP Server
用于与 Neon 管理 API 和数据库交互的 MCP 服务器
Exa MCP Server
模型上下文协议(MCP)服务器允许像 Claude 这样的 AI 助手使用 Exa AI 搜索 API 进行网络搜索。这种设置允许 AI 模型以安全和受控的方式获取实时的网络信息。