Crypto HTTP MCP Server
An HTTP-based Model Context Protocol server that provides tools and resources for accessing real-time cryptocurrency data from the Binance API. It enables users to fetch prices, analyze market changes, and track activity via a streamable endpoint.
README
My HTTP MCP Server
A small Model Context Protocol (MCP) server built with:
@modelcontextprotocol/sdkexpresstypescriptzod
The app exposes a Streamable HTTP MCP endpoint at /mcp and includes a few example tools, resources, and prompts centered around cryptocurrency data.
Project Structure
src/
server.ts
http/
createMcpApp.ts
mcp/
createMcpServer.ts
services/
cryptoService.ts
Responsibilities
-
src/server.tsStarts the HTTP server and wires the app together. -
src/http/createMcpApp.tsContains the HTTP and MCP transport boilerplate: request parsing, origin validation, session management, and/mcproute handling. -
src/mcp/createMcpServer.tsDefines the MCP server itself: tools, resources, prompts, and their schemas. -
src/services/cryptoService.tsContains the business logic for symbol normalization, Binance API calls, and activity log access.
Features
The server currently registers:
-
Tools:
ping,add_numbers,get_price,get_price_change -
Resources:
activity_log,crypto_price -
Prompt:
executive_summary
The crypto-related tools call Binance's public market-data endpoints and append activity to activity.log.
Requirements
- Node.js 20+
- npm
Node 20+ is recommended because the app relies on the built-in fetch API.
Install
npm install
Run
Start the server directly from TypeScript with tsx:
npx tsx src/server.ts
By default, the server listens on:
http://127.0.0.1:8787/mcp
You can override the port with PORT:
PORT=3000 npx tsx src/server.ts
Build / Type Check
This repo does not currently include a tsconfig.json or build script, so the simplest way to validate the TypeScript sources is:
npx tsc --noEmit --module nodenext --moduleResolution nodenext --target es2022 --esModuleInterop src/server.ts src/http/createMcpApp.ts src/mcp/createMcpServer.ts src/services/cryptoService.ts
If you want to add a real build pipeline later, the next step would usually be:
- Add a
tsconfig.json - Add
build,dev, andtypecheckscripts topackage.json - Emit compiled JavaScript into a
dist/directory
Testing
There is not a dedicated automated test suite yet.
Right now, the practical testing flow is:
- Run the TypeScript check command above
- Start the server with
npx tsx src/server.ts - Connect with an MCP client and exercise the registered tools/resources/prompts
Suggested Manual Smoke Tests
- Call
pingand confirm it returnspong - Call
add_numberswith two values and confirm the sum - Call
get_pricewithBTC,ETH,Bitcoin, orEthereum - Call
get_price_changeand confirm a 24-hour summary is returned - Read the
activity_logresource and confirm requests are recorded - Use the
executive_summaryprompt and confirm it references the crypto tools correctly
Notes
- The HTTP boilerplate and session handling are intentionally separate from business logic.
- Session state is stored in memory, so this server is best suited for local development or simple demos.
- The crypto tools depend on outbound network access to Binance.
- The activity log is stored locally in
activity.logat the project root.
Future Improvements
- Add
tsconfig.json - Add npm scripts for
dev,build,typecheck, andtest - Add automated tests for:
- symbol normalization
- Binance response handling
- MCP tool/resource/prompt registration
- HTTP session lifecycle
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。