Sentinel-MCP
Extensible trading-intelligence MCP server with diagnostic tools, structured logging, and tool registration; currently provides server info tool with plans for market data adapters.
README
Sentinel-MCP
Sentinel-MCP is an extensible trading-intelligence platform whose first interface is a Model Context Protocol (MCP) server. Milestone 1 provides a robust server foundation: configuration, structured logging, automatic tool registration, error boundaries, testing, and a diagnostic tool.
Trading execution and market-data integrations are intentionally out of scope for this milestone.
Architecture
The application follows dependency-oriented boundaries:
corecreates the MCP server and cross-cutting concerns such as logging and errors.configvalidates the runtime environment before startup.toolscontains independently registerable MCP capabilities.servicescontains application logic that tools depend on.modelscontains stable, provider-agnostic data contracts.
Future provider adapters (TradingView, Dhan, NSE, news, and caching) can be added behind service interfaces without changing the MCP entry point.
Requirements
- Node.js 22 or later (current LTS recommended)
- pnpm 9 or later
Installation
git clone <your-repository-url>
cd sentinel-mcp
pnpm install
cp .env.example .env
Development
pnpm dev
Build and run
pnpm build
pnpm start
The server communicates over standard input/output, as required by local MCP clients. Logs are written to stderr so they never corrupt the MCP protocol stream.
To use it with Claude Desktop after building, add a server entry that invokes the built file:
{
"mcpServers": {
"sentinel-mcp": {
"command": "node",
"args": ["/absolute/path/to/sentinel-mcp/dist/index.js"]
}
}
}
Available tools
get_server_info
Returns structured runtime data including server name, version, status, ISO timestamp, uptime, Node.js version, and host platform. It is a safe connection and health diagnostic.
Quality commands
pnpm typecheck
pnpm lint
pnpm test
pnpm format:check
Folder structure
src/
config/ Environment validation and runtime settings
core/ Server composition, errors, logging
models/ Provider-agnostic application contracts
providers/ External-provider contracts and future adapters
schemas/ Shared validation schemas
services/ Reusable application services
tools/ Self-registering MCP tools and registry
cache/ Cache implementations and policies
types/ Shared TypeScript declarations
utils/ Small, framework-independent helpers
tests/ Vitest coverage for foundation components
docs/ Project documentation (reserved for design documents)
scripts/ Project automation (reserved for operational scripts)
Configuration
Copy .env.example to .env. The defaults work for local development.
| Variable | Default | Purpose |
|---|---|---|
SENTINEL_APP_NAME |
Sentinel-MCP |
Server identity exposed to clients |
SENTINEL_APP_VERSION |
0.1.0-alpha |
Server version exposed to clients |
SENTINEL_LOG_LEVEL |
info |
Pino log level |
SENTINEL_TRANSPORT |
stdio |
MCP transport for this milestone |
Roadmap
- Provider interfaces and resilient market-data adapters.
- Shared cache, rate-limit, and retry policies.
- Symbol lookup and market intelligence tools.
- News enrichment and configurable alerting.
- Optional authenticated provider integrations.
License
Choose and add an open-source license before public distribution.
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。