apikey-mcp-server
A fully-featured MCP server with API-key authentication, demonstrating realistic e-commerce tools, resources, and prompts.
README
apikey-mcp-server
A fully-featured Model Context Protocol (MCP) server with API-key authentication, demonstrating realistic e-commerce tools, resources, and prompts. Built on the official @modelcontextprotocol/sdk.
Quick Start
npm install
npm run build
# Start with a custom key:
API_KEY=my-secret-key npm start
# Or start without setting API_KEY — uses default key "mcp-api-key-default":
npm start
Authentication
This server requires an API key passed via the x-api-key custom header for HTTP transport access. This is the industry-standard header used by Stripe, SendGrid, GitHub REST API, and many others.
- If
API_KEYenv var is set: use that value as the key. - If
API_KEYenv var is unset: a default keymcp-api-key-defaultis used, so the server is never wide open. - stdio transport: No authentication required — stdio is launched as a local subprocess on a trusted machine.
The /health endpoint is not authenticated so load balancers and monitoring tools can probe it.
Transports
Streamable HTTP
# Start the server
API_KEY=my-secret-key npm start
# Health check (no auth required)
curl http://localhost:3000/health
# Initialize an MCP session
curl -X POST http://localhost:3000/mcp \
-H "Content-Type: application/json" \
-H "x-api-key: my-secret-key" \
-d '{"jsonrpc":"2.0","id":1,"method":"initialize","params":{"protocolVersion":"2026-07-28","capabilities":{},"clientInfo":{"name":"test","version":"1.0.0"}}}'
stdio
npm run start:stdio
# Or: npx @modelcontextprotocol/inspector node dist/stdio.js
MCP Inspector
- Transport:
Streamable HTTP - URL:
http://localhost:3000/mcp - Custom Header:
x-api-key = your-key-here
Configuration
| Variable | Default | Description |
|---|---|---|
API_KEY |
mcp-api-key-default |
API key for HTTP auth. Always falls back to this default if unset. |
PORT |
3000 |
HTTP server port |
HOST |
127.0.0.1 |
Bind address (auto-switches to 0.0.0.0 when K_SERVICE set — Cloud Run) |
NODE_ENV |
development |
Environment name |
MCP_LOG_BODY_LIMIT |
4000 |
Max chars of request/response body to log |
MCP Capabilities
Tools (7)
| Tool | Description |
|---|---|
search_customers |
Search 25 customer records by name, email, or ID |
get_order |
Fetch full order details with line items, pricing, and tracking |
list_products |
Browse 30 products across 5 categories with text search and category filters |
calculate_shipping |
Estimate shipping cost by product weight and destination zone |
generate_report |
Generate business reports: sales, customers, orders, inventory, or summary |
validate_address |
Validate and standardize postal addresses with confidence scoring |
ping |
Connectivity and health check with uptime info |
Resources (5)
| URI | Description |
|---|---|
config://app |
Server configuration and feature flags (JSON) |
catalog://products |
Full product catalog — 30 products (JSON) |
docs://api |
API documentation listing all capabilities (Markdown) |
customers://{id} |
Individual customer record with order history (JSON) |
orders://{id} |
Individual order record with line items and tracking (JSON) |
Prompts (4)
| Prompt | Description |
|---|---|
summarize_customer |
Generate a customer summary for support staff |
draft_order_email |
Draft an order status email to a customer |
product_recommendation |
Generate personalized product recommendations |
data_insights |
Analyze business metrics and surface actionable insights |
Mock Data
All data is in-memory (no external database required):
- 25 customers across 4 tiers (bronze / silver / gold / platinum)
- 30 products across 5 categories (Electronics, Apparel, Home, Books, Sports)
- 30 orders across 5 statuses (pending, confirmed, shipped, delivered, cancelled)
- 4 shipping zones with weight-tiered rate tables
MCP Protocol Compatibility
| Spec | Transport | Status |
|---|---|---|
| 2026-07-28 | Streamable HTTP | ✅ Primary |
| 2025-11-25 | HTTP + SSE | ✅ Compatible |
Claude Desktop Configuration
Add to claude_desktop_config.json:
{
"mcpServers": {
"apikey-mcp-server": {
"command": "node",
"args": ["C:\\Users\\...\\apikey-mcp-server\\dist\\stdio.js"]
}
}
}
Docker
docker build -t apikey-mcp-server .
docker run -p 8080:8080 -e API_KEY=my-secret-key apikey-mcp-server
Scripts
| Script | Description |
|---|---|
npm run build |
Compile TypeScript to dist/ |
npm start |
Start HTTP server |
npm run start:stdio |
Start stdio server |
npm run dev |
Watch mode (TypeScript compilation only) |
npm run dev:http |
Hot-reload HTTP server with tsx |
npm run dev:stdio |
Hot-reload stdio server with tsx |
npm run inspect |
Launch MCP Inspector (stdio) |
npm run test:smoke |
Build + run smoke test |
npm run clean |
Remove dist/ |
Project Structure
src/
├── server.ts # MCP server factory (createServer)
├── http.ts # Streamable HTTP transport + auth middleware
├── stdio.ts # Stdio transport entry point
├── auth.ts # API key validation (x-api-key header)
├── config.ts # Typed configuration loader (env vars)
├── data.ts # Mock data store (customers, products, orders, shipping)
├── tools.ts # 7 e-commerce tools
├── resources.ts # 5 resources (static + templated)
├── prompts.ts # 4 prompts
├── banner.ts # ANSI startup banner
└── logging.ts # Structured JSON logging with redaction
scripts/
└── smoke.mjs # stdio smoke test
License
MIT
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。