mcp-server-indicators

carlosatta/mcp-server-indicators

3.2

If you are the rightful owner of mcp-server-indicators and would like to certify it and/or have it hosted online, please leave a comment on the right or send an email to dayong@mcphub.com.

The Trading Indicators Server is a specialized MCP server that provides comprehensive technical analysis indicators for trading applications using the Tulind library.

Tools
8
Resources
0
Prompts
0

MCP Server - Trading Indicators

🔧 Trading Indicators Server: A specialized MCP server providing comprehensive technical analysis indicators for trading applications.

A MCP (Model Context Protocol) server that exposes trading indicators via Server-Sent Events (SSE) transport. This server provides access to popular technical analysis indicators using the powerful Tulind library.

Features

  • 🌐 Web-based MCP server using SSE transport
  • 📊 7 Popular Trading Indicators: RSI, EMA, SMA, MACD, Bollinger Bands, Stochastic, ATR
  • �� Extensible tool system for adding custom indicators
  • 🔐 Environment-based configuration management
  • 📝 Comprehensive logging for debugging
  • 🔄 Session-based transport with UUID tracking
  • 🛠️ Modular architecture for easy customization
  • ⚡ High Performance using Tulind (C-based) library
  • 🔍 Automatic Port Detection - finds available ports automatically

Trading Indicators Available

Momentum Indicators

  • RSI (Relative Strength Index) - Identify overbought/oversold conditions
  • Stochastic Oscillator - Compare closing price to price range
  • MACD (Moving Average Convergence Divergence) - Trend and momentum

Trend Indicators

  • EMA (Exponential Moving Average) - Trend following with recent price emphasis
  • SMA (Simple Moving Average) - Classic trend following indicator

Volatility Indicators

  • ATR (Average True Range) - Measure market volatility
  • Bollinger Bands - Volatility bands around moving average

Installation

npm install

Configuration

Create a .env file in the root directory:

# Server Configuration
HOST=0.0.0.0
PORT=3000
LOG_LEVEL=info

# CORS Configuration (optional)
# CORS_ORIGINS=http://localhost:3000,http://localhost:8080

Running the Server

npm start

The server will automatically find an available port if 3000 is occupied.

Available Tools

Basic Tools

  • get_server_info - Server information including available indicators

Trading Indicators

1. RSI (Relative Strength Index)
{
  "name": "calculate_rsi",
  "arguments": {
    "prices": [100, 102, 101, 103, 105, 104, 106],
    "period": 14
  }
}
2. EMA (Exponential Moving Average)
{
  "name": "calculate_ema",
  "arguments": {
    "prices": [100, 102, 101, 103, 105, 104, 106],
    "period": 20
  }
}
3. SMA (Simple Moving Average)
{
  "name": "calculate_sma",
  "arguments": {
    "prices": [100, 102, 101, 103, 105, 104, 106],
    "period": 20
  }
}
4. MACD
{
  "name": "calculate_macd",
  "arguments": {
    "prices": [100, 102, 101, 103, 105, 104, 106],
    "fastPeriod": 12,
    "slowPeriod": 26,
    "signalPeriod": 9
  }
}
5. Bollinger Bands
{
  "name": "calculate_bollinger_bands",
  "arguments": {
    "prices": [100, 102, 101, 103, 105, 104, 106],
    "period": 20,
    "stdDev": 2.0
  }
}
6. Stochastic Oscillator
{
  "name": "calculate_stochastic",
  "arguments": {
    "high": [105, 107, 106, 108, 110, 109, 111],
    "low": [95, 97, 96, 98, 100, 99, 101],
    "close": [100, 102, 101, 103, 105, 104, 106],
    "kPeriod": 14,
    "kSmoothPeriod": 3,
    "dPeriod": 3
  }
}
7. ATR (Average True Range)
{
  "name": "calculate_atr",
  "arguments": {
    "high": [105, 107, 106, 108, 110, 109, 111],
    "low": [95, 97, 96, 98, 100, 99, 101],
    "close": [100, 102, 101, 103, 105, 104, 106],
    "period": 14
  }
}

Integration with CCXT

These indicators are designed to work seamlessly with CCXT OHLCV data:

// Example: Get BTC/USDT data from exchange and calculate RSI
const ohlcv = await exchange.fetchOHLCV('BTC/USDT', '1h', undefined, 100);
const closes = ohlcv.map(candle => candle[4]); // Extract close prices

// Use with MCP client
const rsiResult = await client.callTool({
  name: "calculate_rsi",
  arguments: {
    prices: closes,
    period: 14
  }
});

Data Format

Input Data

  • prices: Array of closing prices (numbers)
  • high/low/close: Arrays for OHLC indicators (all same length)
  • period: Integer for calculation period

Output Format

All indicators return structured JSON with:

  • indicator: Name of the indicator
  • values: Calculated indicator values
  • current: Current/latest values with interpretation
  • analysis: Additional analysis and signals
  • parameters: Used parameters

Testing

node test-mcp-client.js

Architecture

src/
├── tools/
│   ├── tradingIndicators.js    # Index of all indicators
│   ├── rsi.js                  # RSI indicator
│   ├── ema.js                  # EMA indicator
│   ├── sma.js                  # SMA indicator
│   ├── macd.js                 # MACD indicator
│   ├── bollingerBands.js       # Bollinger Bands
│   ├── stochastic.js           # Stochastic Oscillator
│   ├── atr.js                  # ATR indicator
│   └── publicTools.js          # Public tools including indicators
├── config/
│   └── config.js               # Server configuration
└── mcpServer.js                # MCP server logic

Adding Custom Indicators

  1. Create a new file in src/tools/ (e.g., myIndicator.js)
  2. Export myIndicatorDefinition and myIndicatorHandler
  3. Add to src/tools/tradingIndicators.js
  4. Restart the server

Performance

  • Tulind Library: C-based calculations for maximum performance
  • Efficient Memory Usage: Streaming calculations where possible
  • Error Handling: Comprehensive validation and error messages
  • Parallel Execution: Multiple indicators calculated concurrently (1-5ms typical)
  • No Event Loop Blocking: process.nextTick protection for tulind operations

MCP Standard Compliance

This server follows the MCP specification revision 2025-06-18:

Session Management

  • ✅ HTTP 404 (-32004) for unknown session IDs
  • ✅ HTTP 400 (-32602) for missing session IDs
  • ✅ 5-minute session timeout (configurable via MCP_SESSION_TIMEOUT_MS)
  • ✅ 30-second cleanup interval (configurable via MCP_CLEANUP_INTERVAL_MS)
  • ✅ Graceful session cleanup on timeout

Error Handling

  • ✅ Structured error responses (no thrown exceptions)
  • ✅ Transport blocking prevention
  • ✅ Individual operation timeouts (1s per indicator)
  • ✅ Global operation timeout (5s for batch calculations)
  • ✅ Partial results on timeout (trading system protection)

Error Codes

  • -32602: Invalid parameters / Session ID required
  • -32004: Session not found
  • -32000: Bad request
  • -32603: Internal server error
  • -32601: Method not found

Testing

Run conformance tests to verify MCP compliance:

# Basic conformance test
npm run test

# Extended conformance test (session lifecycle, errors, parallel execution)
node test-extended-conformance.js

# Stress test (concurrent execution, timeout scenarios)
node test-stress-cascade.js

Configuration

Environment variables for MCP behavior:

# Session timeout (default: 5 minutes)
MCP_SESSION_TIMEOUT_MS=300000

# Cleanup interval (default: 30 seconds)
MCP_CLEANUP_INTERVAL_MS=30000

# Tool execution timeout (default: 20 seconds)
TOOL_EXECUTION_TIMEOUT_MS=20000

# Allow auto-recreate sessions (default: false, strict compliance)
ALLOW_AUTO_SESSION_RECREATE=false

# Enable session debug endpoint (default: auto-detect)
ENABLE_SESSION_DEBUG=true

Use Cases

  • Algorithmic Trading: Real-time indicator calculations
  • Portfolio Analysis: Technical analysis of holdings
  • Market Research: Backtesting and strategy development
  • Risk Management: Volatility and trend analysis
  • AI/ML Training: Feature generation for models

Architecture Notes

Transport Blocking Prevention

This implementation includes protection against transport blocking cascade failures:

  • Aggressive timeouts (1s per indicator, 5s total)
  • Session cleanup on timeout
  • Structured error responses (never throws exceptions)
  • Parallel indicator execution
  • Event loop protection via process.nextTick

Legacy Files

The legacy/ directory contains deprecated implementations for reference only. Do not use these files in production.

Contributing

  1. Fork the repository
  2. Create your feature branch (git checkout -b feature/new-indicator)
  3. Add your indicator following the existing pattern
  4. Test your changes
  5. Submit a pull request

License

This project is licensed under the ISC License.