sentimatix-mcp 1.0.0

Model Context Protocol server for Sentimatix - Indian Stock Market Sentiment Intelligence

Sentimatix MCP Server

Real-time Indian stock market intelligence via the Model Context Protocol (MCP). Plug Sentimatix directly into Claude Desktop, Cursor, or any MCP-compatible AI agent to get live NSE/BSE sentiment, news, and technical analysis.

🏗️ Architecture

User Query → Claude/AI Agent → MCP Server → Tools
                                   ↓
                              ┌─────────────────────┐
                              │ News Sentiment API  │ ← Sentimatix /api/v1/news
                              │ Stock Price API     │ ← Sentimatix /api/v1/stocks
                              │ Sector Sentiment    │ ← Sentimatix /api/v1/sentiment
                              │ RAG Evidence        │ ← Supabase Vector DB
                              └─────────────────────┘
                                   ↓
                              Synthesized Market Intelligence

🛠️ Available Tools (10 total)

Tool	Description	Example
explain_price_change	Orchestrator — explains why a stock moved using price, news + sentiment	(RELIANCE, 7 days)
analyze_stock_enhanced	Deep single-stock research report	(HDFCBANK, detailed)
compare_stocks	Side-by-side comparison of two NSE stocks	(TCS vs INFY)
get_stock_summary	Price metrics: change%, high, low, volume	(TATAMOTORS, 7 days)
get_historical_prices	Daily OHLCV time-series data	(SBIN, 2024-01-01, 2024-03-01)
get_news_sentiment	News articles with NLP sentiment scores	(ZOMATO, last 7 days)
get_sentiment_aggregate	Aggregated sentiment stats for a period	(ADANIENT, 30 days)
get_technical_analysis	RSI, MACD, Bollinger Bands, support/resistance	(ITC, 90 days)
calculate_correlation	Pearson correlation between two NSE stocks	(RELIANCE, TCS)
get_rag_evidence	Semantic search over the news corpus	(INFY, "earnings beat")

🚀 Quick Start (Claude Desktop)

Add this to your claude_desktop_config.json:

{
  "mcpServers": {
    "sentimatix": {
      "command": "python",
      "args": ["path/to/apps/mcp/mcp_stdio.py"],
      "env": {
        "BACKEND_API_URL": "https://sentimatix-production.up.railway.app/api"
      }
    }
  }
}

No API key is required for the default free tier.

📊 Example Query Flow

User: "Why did Reliance drop in the last 7 days?"

MCP Workflow:

1. ✅ get_stock_summary("RELIANCE", 7) → -3.1% decline
2. ✅ get_news_sentiment("RELIANCE", ...) → 12 news items, avg sentiment: -0.4
3. ✅ get_rag_evidence("RELIANCE", "price drop reasons") → 5 relevant articles
4. ✅ get_technical_analysis("RELIANCE") → RSI: 38 (Oversold), MACD: Bearish
5. 🤖 LLM Synthesis:

RELIANCE declined 3.1% (₹2,850 → ₹2,762) over the last 7 days.

Top Reasons (with evidence):
1. Weak Q4 refining margins — reported by Economic Times (sentiment: -0.5)
2. Sector-wide FII selling in Oil & Gas — MoneyControl (sentiment: -0.3)
3. RSI at 38 confirms oversold territory; key support at ₹2,740

Confidence: HIGH

🔧 Environment Variables

BACKEND_API_URL=https://sentimatix-production.up.railway.app/api
SUPABASE_URL=https://your-project.supabase.co       # For RAG features
SUPABASE_KEY=your_anon_key                           # For RAG features
OPENAI_API_KEY=your_openai_key                       # For AI synthesis

📚 Full Documentation

• Connection Guide — Claude Desktop, Cursor setup
• Quick Start — Step-by-step installation
• REST API Docs — Full OpenAPI reference