Guys, stock data no longer needs a backend! chengzuopeng/stock-sdk —— Pure JS/TS stock market SDK, zero Python, zero backend, get real-time data directly from browser/Node.js!
Key highlights:

• Supports A-shares, H-shares, US stocks, public mutual funds – real-time quotes, K-line, technical indicators all included
• Namespaced API is super friendly, with IDE autocomplete
• Built-in CLI and MCP – one command to connect to AI tools
• Lightweight, zero dependencies, perfect TypeScript support
  Want to build a market dashboard, quant demo, or AI stock picker? Just npm i stock-sdk and start!
  Direct link: https://github.com/chengzuopeng/stock-sdk
  Official docs: https://stock-sdk.linkdiary.cn
  Quant and frontend developers, go for it – this open source benefit is really practical~
  #StockSDK #StockAPI #FrontendQuant #JSStockData

chengzuopeng/stock-sdk

Source: https://github.com/chengzuopeng/stock-sdk

Stock SDK

npm version | npm downloads | license | MCP | AI Ready

English | 中文

A JavaScript SDK for stock market quotes designed for frontend and Node.js.

No Python, no backend service required – get real-time quotes and K-line data for A-shares / H-shares / US stocks / public mutual funds directly in browser or Node.js. Also comes with a command-line tool and MCP server – one command to fetch quotes or integrate with AI.

✨ Zero dependency, lightweight package | 🌐 Browser + Node.js | 📦 ESM + CJS + subpath | 🧠 Full TypeScript types | 🖥️ CLI | 🤖 MCP

✨ v2.0.0: v2 is an architectural leap (namespaced API, unified symbol model, Quote discriminated union, unified error system, CLI/MCP/subpath exports).
Installation: npm i stock-sdk. To upgrade from v1, first read the v1 → v2 migration guide (breaking changes, no compatibility aliases).

📖 Official Documentation

👉 https://stock-sdk.linkdiary.cn

Complete API, namespace overview, CLI/MCP guides, online Playground, v1 → v2 migration docs are all here. Check the site first before diving in.

(v1 stable docs v1.stock-sdk.linkdiary.cn are archived)

📦 NPM | 📖 GitHub | 🎮 Online Playground

🧭 Stock Dashboard: A stock data dashboard demo site built on stock-sdk – welcome to try it.

Why stock-sdk?

If you’re a frontend engineer, you may have encountered these problems:

• Most stock market tools are Python-based, making them hard to use directly in frontend
• You want to build a market dashboard/demo but don’t want to maintain an extra backend service
• Financial API response formats are messy and encoding is complex (GBK/concurrency/batch)

The goal of stock-sdk is simple:

Let frontend engineers elegantly fetch stock market data using the JavaScript/TypeScript they know best.

Use Cases

• 📊 Stock market dashboard (Web/Admin)
• 📈 Data visualization (ECharts/TradingView)
• 🎓 Stock/finance course demos
• 🧪 Quant strategy prototyping (JS/Node)
• 🕒 Scheduled stock data fetching with Node.js
• 🖥️ Quick quote lookup via CLI / 🤖 Feed data source to AI tools

Features

• ✅ Zero dependencies, runs on both browser and Node.js 18+; provides ESM and CommonJS
• ✅ Namespaced API: sdk.quotes.cn() / sdk.kline.cn() / sdk.options.etf.dailyKline(), grouped by domain, IDE autocomplete friendly
• ✅ Unified symbol model: string as first-class citizen – tolerant parsing of sh600519, 600519, 600519.SH, 00700, hk00700, AAPL, 105.AAPL etc.; supports special indices like CSI (930955, H30533, HSHCI, GDAXI – see Symbol Guide)
• ✅ A-shares / H-shares / US stocks / public mutual funds real-time quotes, historical K-line (daily/weekly/monthly), minute K-line (1/5/15/30/60), intraday time-sharing
• ✅ Technical indicators: MA, MACD, BOLL, KDJ, RSI, WR, BIAS, CCI, ATR, OBV, ROC, DMI, SAR, KC
• ✅ Signals / Screener / Backtest: calcSignals (golden cross/death cross/overbought/oversold event recognition), chainable screener, local backtest
• ✅ Futures / Options / Capital flow / Dragon & Tiger list / Northbound / Block trades / Margin trading / Limit-up board – full extended data
• ✅ Mutual fund deep data: historical NAV, real-time estimate, peer ranking trend, fund/ETF dividends, theme funds
• ✅ Subpath exports: stock-sdk/{indicators,signals,symbols,screener,cache,errors} – pure computation without network layer, tree-shake friendly
• ✅ Unified error system: only throws SdkError with a standard code, importable from stock-sdk/errors
• ✅ Request governance: provider-level retry/rate-limit/circuit-breaker + injectable fetchImpl/signal/lifecycle hooks
• ✅ CLI: stock-sdk quote 600519 – fetch quotes directly from terminal
• ✅ Built-in MCP server: stock-sdk mcp – one line to integrate with Cursor/Claude/Codex and other AI tools (zero-dependency hand-written, no @modelcontextprotocol/sdk)

Installation

# Latest version (v2: namespaced API / CLI / MCP)  
npm install stock-sdk  

# v1 legacy (archived, critical fixes only)  
npm install stock-sdk@legacy  

Quick Start

import { StockSDK } from 'stock-sdk';  

const sdk = new StockSDK();  

// Namespaced API (v2) – symbol format tolerant, '600519' / 'sh600519' / '600519.SH' all work  
const quotes = await sdk.quotes.cnSimple(['sh000001', 'sz000858', 'sh600519']);  
quotes.forEach((q) => {  
  console.log(`${q.name}: ${q.price} (${q.changePercent}%)`);  
});  

// Historical K-line + technical indicators  
const kline = await sdk.kline.withIndicators('600519', {  
  period: 'daily',  
  indicators: {  
    ma: { periods: [5, 10, 20] },  
    macd: {},  
  },  
});  

// All A-share quotes (5000+ stocks, built-in concurrency control)  
const all = await sdk.batch.cn({ concurrency: 5 });  
console.log(`Total ${all.length} stocks`);  

H-shares '00700' / 'hk00700', US stocks 'AAPL' / '105.AAPL' – all handled by normalizeSymbol with unified tolerant parsing.

CLI

After installation, the stock-sdk command is available (or use npx):

# Quote (auto-detect market by code)  
npx stock-sdk quote 600519 00700 AAPL  

# K-line + truncated output  
npx stock-sdk kline 600519 --period weekly --limit 30  

# With technical indicators  
npx stock-sdk indicators 600519 --ma 5,10,20 --macd  

# Search  
npx stock-sdk search 茅台  

# Directly call any namespace method  
npx stock-sdk quotes cn sh600519 sz000001  

Default JSON output, add --format table|csv, --pretty, --limit N.

🤖 AI / MCP Integration

v2 includes a built-in zero-dependency MCP server – start with one command:

npx stock-sdk mcp  

Connect to Cursor / Claude Desktop / Codex / Gemini etc. (configure mcpServers):

{  
  "mcpServers": {  
    "stock-sdk": {  
      "command": "npx",  
      "args": ["-y", "stock-sdk", "mcp"]  
    }  
  }  
}  

Environment variable STOCK_SDK_MCP_TOOLS=core|full|<comma-separated tool names> controls the tool set (default core).

👉 Full MCP Documentation

Signals / Screener / Backtest

import { calcSignals } from 'stock-sdk/signals';  
import { screen, backtest } from 'stock-sdk/screener';  

// Event recognition (golden cross/death cross/overbought/oversold) based on K-line with indicators  
const signals = calcSignals(klineWithIndicators, {  
  ma: { fast: 5, slow: 20 },  
  rsi: {},  
});  

// Chainable screener (input any quote array, local only, no network)  
const picks = screen(allQuotes)  
  .where((q) => q.pe != null && q.pe < 20)  
  .where((q) => q.changePercent > 3)  
  .sortBy((q) => q.amount)  
  .top(20);  

// Local backtest  
const report = backtest({  
  klines,  
  strategy: (bar, i, all) => 'hold', // returns 'buy' | 'sell' | 'hold'  
});  
console.log(report.totalReturn, report.winRate, report.maxDrawdown);  

Request Governance & Errors

import { StockSDK } from 'stock-sdk';  
import { HttpError, getSdkErrorCode } from 'stock-sdk/errors';  

const sdk = new StockSDK({  
  retry: { maxRetries: 2, baseDelay: 500 },  
  providerPolicies: {  
    eastmoney: { timeout: 12000, rateLimit: { requestsPerSecond: 3, maxBurst: 3 } },  
  },  
});  

try {  
  await sdk.quotes.cnSimple(['sh600519']);  
} catch (error) {  
  // v2 only throws SdkError with unified code  
  if (error instanceof HttpError) console.log(error.status, error.statusText);  
  console.log(getSdkErrorCode(error)); // HTTP_ERROR / NETWORK_ERROR / TIMEOUT / ABORTED / PARSE_ERROR ...  
}  

Subpath Exports

When using only pure calculations (indicators/symbols/signals/screener), import from subpaths to avoid pulling in RequestClient and all providers:

// 14 indicators, 17 pure functions (MA family includes calcSMA, calcEMA, calcWMA):  
// calcSMA / calcEMA / calcWMA / calcMA / calcMACD / calcBOLL / calcKDJ /  
// calcRSI / calcWR / calcBIAS / calcCCI / calcATR / calcOBV / calcROC /  
// calcDMI / calcSAR / calcKC  
import { calcMACD, calcKDJ } from 'stock-sdk/indicators';  
import { normalizeSymbol, toTencentSymbol } from 'stock-sdk/symbols';  
import { calcSignals } from 'stock-sdk/signals';  
import { screen, backtest } from 'stock-sdk/screener';  
import { MemoryCacheStore, cacheThrough } from 'stock-sdk/cache';  
import { SdkError, isSdkError, getSdkErrorCode } from 'stock-sdk/errors';  

Market Support Matrix

Different markets have varying levels of coverage. The table below helps you quickly determine if the SDK covers your scenario.

• ✅ Supported | ⚠️ Partial (see notes) | ❌ Not implemented | — Not applicable

Data delay: Real-time quotes come from public sources like Tencent Finance / East Money – not real-time matching, usually tens of seconds to minutes delay. Not suitable for high-frequency trading decisions.

API Overview (Namespaces)

💡 Full API in Official Documentation. In v2, all methods are under namespaces:

Indicator calculations moved to subpath: import { calcMACD } from 'stock-sdk/indicators'.
Migrating from v1 flat API? See v1 → v2 Migration Guide (includes full sdk.getXxx() → sdk.<namespace>.<method>() mapping table).

Development Checks

pnpm typecheck  
pnpm build  
pnpm test  
pnpm test:integration:smoke   # Smoke integration (real network)  
pnpm test:integration:full    # Full integration regression  

License

ISC

🌐 Official Site | 📦 NPM | 📖 GitHub | 🎮 Online Demo | 🧭 Stock Dashboard | 🐛 Issues

If this project helps you, feel free to Star ⭐ or raise an Issue for feedback.