siftingio-mcp

siftingio-mcp

MCP server that provides access to SiftingIO market data, including live prices, SEC filings, OHLCV bars, 13F holdings, market status, and economic calendar tools for AI assistants.

Category
访问服务器

README

siftingio-mcp

This is a Model Context Protocol (MCP) server that puts the SiftingIO market-data SDK (@siftingio/sdk) in reach of your AI assistant. Once it's running, the model can pull live prices, dig through SEC/EDGAR fundamentals, fetch OHLCV bars, look up 13F holdings, check market status, and scan the macro economic calendar — all as tools.

Setup

npm install
npm run build

You'll need an API key, which you can grab at https://sifting.io:

export SIFTING_API_KEY=sft_...

If you need to point at a different backend, SIFTING_BASE_URL and SIFTING_WS_URL are there to override the defaults.

Working locally? Copy .env.example to .env instead — npm run dev and npm start pick it up automatically (Node does the loading via --env-file-if-exists). When you wire this into an actual MCP client, though, pass the key through the server's env block rather than a file (there's an example further down).

Run

Here's the toolbox:

  • npm run build — compile TypeScript into dist/.
  • npm start — run the compiled server (node dist/index.js) over stdio.
  • npm run start:http — run it over Streamable HTTP (node dist/http.js).
  • npm run dev / npm run dev:http — run straight from source with tsx, no build step.
  • npm test — run the vitest suite (add npm run test:watch to keep it running).
  • npm run lint / npm run format — ESLint (typescript-eslint) and Prettier.
  • npm run typechecktsc --noEmit.

On every push and PR, CI (.github/workflows/ci.yml) walks through the same gauntlet: format check → lint → typecheck → build → test.

One thing worth knowing: the server talks JSON-RPC over stdio, so stdout belongs entirely to the protocol. Anything diagnostic goes to stderr to stay out of the way.

Inspect interactively

Want to poke at it by hand? The MCP inspector is the easiest way:

SIFTING_API_KEY=sft_... npx @modelcontextprotocol/inspector node dist/index.js

HTTP (Streamable HTTP) transport

If you're running this somewhere remote or hosted, use MCP's Streamable HTTP transport instead of stdio:

SIFTING_API_KEY=sft_... PORT=3000 npm run start:http
# → MCP endpoint at http://127.0.0.1:3000/mcp  (POST messages, GET SSE, DELETE session)

It's stateful: every client gets its own session (tracked by the mcp-session-id header) and its own McpServer, while the upstream SiftingIO connection is shared across the whole process. As a safety measure it only binds to loopback and turns away non-local browser Origins — that's the DNS-rebinding protection. Set the port with PORT (or MCP_HTTP_PORT); it defaults to 3000.

If you want auth, set MCP_AUTH_TOKEN and the server will demand Authorization: Bearer <token> on every request (anything missing or wrong gets a 401). Pair that with a reverse proxy handling TLS and the token, and you can safely expose the server past localhost:

MCP_AUTH_TOKEN=s3cret SIFTING_API_KEY=sft_... npm run start:http

Then just point any HTTP-capable MCP client at http://127.0.0.1:3000/mcp:

claude mcp add --transport http siftingio http://127.0.0.1:3000/mcp

Use with an MCP client

Drop this into your client config — Claude Desktop's claude_desktop_config.json, say, or use claude mcp add if you're on Claude Code:

{
  "mcpServers": {
    "siftingio": {
      "command": "node",
      "args": ["/absolute/path/to/siftingio-mcp/dist/index.js"],
      "env": { "SIFTING_API_KEY": "sft_..." }
    }
  }
}

Tools (36)

Namespace Tools
Live (snapshot) last_trade, last_quote, last_tvl
Stocks stocks_search, stocks_profile, stocks_filings, stocks_filing, stocks_sections, stocks_section, stocks_risk_factors_diff, stocks_ratios, stocks_earnings, stocks_financials, stocks_financial_concept, stocks_insiders, stocks_ownership, stocks_events, stocks_compensation, stocks_screener, stocks_bars
Crypto / Forex crypto_bars, forex_bars
DEX dex_wallet
Markets markets_list, markets_status_all, markets_status, markets_hours, markets_calendar
Filers filers_holdings
Macro economic_calendar_list
Live (stream) ws_subscribe, ws_unsubscribe, ws_poll, ws_collect, ws_status, ws_disconnect

A few patterns are worth calling out:

Paginated tools take cursor/limit and hand back a meta.next_cursor to fetch the next page. The stocks_* list tools — stocks_filings, stocks_earnings, stocks_insiders, stocks_ownership, stocks_events, stocks_compensation — also understand max_items: set it and they'll auto-paginate, gathering up to that many items across pages in a single call.

The high-traffic tools (last_trade, last_quote, last_tvl, stocks_profile, stocks_search) come with an output schema and return structuredContent alongside the human-readable text, so clients can read them machine-side too.

Every tool also carries MCP annotations. The data tools are readOnlyHint: true (and openWorldHint: true, since they reach out to the external API), while the WebSocket tools that change connection state are readOnlyHint: false, destructiveHint: false.

Results are size-capped at roughly 60k characters (see MAX_RESULT_CHARS in src/util.ts). When a heavy endpoint — full XBRL financials, screeners, OHLCV bars — returns more than that, the server trims its largest array and tacks on a _truncated note explaining how to narrow the query.

Live WebSocket streaming

Streaming is the awkward case: it doesn't fit neatly into a single request/response. So the server holds one persistent WebSocket open, buffers the frames as they arrive, and the tools just read from that buffer. Channels (the product field) are cex (crypto), dex (DEX trades), fx (forex), us (US stocks), and tvl (DEX pool TVL).

There are two ways to work with it:

  • Subscribe + poll, for ongoing streams: call ws_subscribe once, then keep calling ws_poll. The first poll gives you a recent tail; feed the returned next_seq back in as after_seq and you'll only get newer frames from then on. ws_status shows you the connection and what's subscribed, and ws_disconnect tears the whole thing down.
  • Collect, for a quick one-shot: ws_collect subscribes, waits up to duration_ms (or until it's seen max frames), returns what it caught, and cleans up any subscription it had to create. Perfect for "grab me a few seconds of BTCUSD."

The connection reconnects on its own and replays your subscriptions when it does. The buffer is a rolling window, so the oldest frames eventually fall off — and when they do, you'll hear about it through dropped/gap.

Prompts

These are guided, multi-tool workflows your client can surface as slash-commands:

  • company_snapshot (ticker) — pulls stocks_profile, stocks_ratios, the latest stocks_filings, and last_trade together into one briefing.
  • compare_companies (tickers) — lines up several tickers side by side across the key ratios.
  • market_now — what's open and closed right now, plus the high-impact macro events coming up.

Logging & shutdown

The server advertises the MCP logging capability and pushes structured notifications/message to the client whenever something happens with the connection (WebSocket open/close/reconnect/error) or on shutdown — and it mirrors all of that to stderr too.

When it catches a SIGINT or SIGTERM, it closes the live WebSocket and shuts the server down cleanly before exiting.

推荐服务器

Baidu Map

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。

官方
精选
JavaScript
Playwright MCP Server

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。

官方
精选
TypeScript
Magic Component Platform (MCP)

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。

官方
精选
本地
TypeScript
Audiense Insights MCP Server

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。

官方
精选
本地
TypeScript
VeyraX

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。

官方
精选
本地
Kagi MCP Server

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。

官方
精选
Python
graphlit-mcp-server

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。

官方
精选
TypeScript
e2b-mcp-server

e2b-mcp-server

使用 MCP 通过 e2b 运行代码。

官方
精选
Neon MCP Server

Neon MCP Server

用于与 Neon 管理 API 和数据库交互的 MCP 服务器

官方
精选
Exa MCP Server

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。

官方
精选