deepansh-saxena/Stocks-Analyzer-AI

Stock Advisor Agent MCP

A sophisticated stock market analysis assistant that combines real-time stock data, news analysis, and AI-powered recommendations.

Features

• Real-time Stock Data: Fetches stock information from Zacks quote feed
• News Analysis: Gets latest news headlines using NewsAPI
• AI-Powered Analysis: Uses GPT-4 to analyze stock data and news sentiment
• Web Interface: Modern React frontend with markdown formatting
• Contextual Chat: Maintains conversation history for follow-up questions
• MCP Integration: Uses Model Context Protocol for tool integration

Architecture

Frontend (React) → API Server (Flask) → MCP Client → Stocks Server (MCP Tools)

Setup Instructions

1. Create Virtual Environment (Recommended)

# Create virtual environment
python -m venv venv

# Activate virtual environment
# On macOS/Linux:
source venv/bin/activate
# On Windows:
# venv\Scripts\activate

2. Install Dependencies

# Install Python dependencies
pip install mcp langchain-mcp-adapters langgraph openai flask flask-cors requests newsapi-python

# Install frontend dependencies
cd frontend
npm install

3. Configure API Keys

Update the API keys in the following files:

• api_server.py: Update your OpenAI API key (replace #OPENAI_API_KEY with your actual key)
• stocks_server.py: Update your NewsAPI key (replace "Api Key Here" with your actual key)

4. Start the Servers

# Terminal 1: Start API server (this will automatically start the stocks server as needed)
python api_server.py

# Terminal 2: Start frontend
cd frontend
npm run dev

Note: You do NOT need to manually start stocks_server.py. It is automatically launched as a subprocess by the API server when needed.

5. Access the Application

• Frontend: http://localhost:5173 (or the port shown by Vite)
• API Server: http://localhost:5050
• Health Check: http://localhost:5050/api/health

Usage

1. Open the frontend in your browser
2. Ask questions about stocks, for example:
   • "How is Microsoft looking today?"
   • "Should I buy Apple stock?"
   • "Get news about Tesla and analyze the stock"
   • "What's the latest on NVIDIA?"

The AI agent will:

• Fetch real-time stock data from Zacks
• Get relevant news headlines from NewsAPI
• Analyze the information using GPT-4
• Provide a BUY/HOLD/SELL recommendation with detailed reasoning
• Format responses with markdown tables, bold text, and clear structure

API Endpoints

• POST /api/analyze: Analyze stock based on user message
• GET /api/health: Health check endpoint

Tools Available

• get_stock_data(ticker): Fetches stock data from Zacks quote feed
• get_news(keyword, category, language, country): Gets news articles from NewsAPI

Features

Contextual Chat

• Maintains conversation history per session
• Follow-up questions reference previous analysis
• Avoids repeating information unnecessarily

Rich Formatting

• Markdown tables for key metrics
• Bold text for important numbers and headers
• Clickable news links
• Blockquotes for recommendations
• Professional but conversational tone

Real-time Analysis

• Current stock prices and performance
• Analyst ratings and recommendations
• Financial metrics (P/E ratio, EPS, revenue growth)
• Technical indicators and trends
• Recent news sentiment analysis

Troubleshooting

Dependencies Issues

• Ensure you're in the virtual environment: source venv/bin/activate
• Install all required packages: pip install -r requirements.txt (if available)
• For frontend issues: cd frontend && npm install

API Connection Issues

• Verify the API server is running on port 5050
• Check CORS settings if frontend can't connect
• Ensure OpenAI API key is set in api_server.py

News API Issues

• Verify your NewsAPI key is valid and set in stocks_server.py
• Check API quota limits
• Ensure the keyword is not too specific

Stocks Server Issues

• The stocks server runs as a subprocess - no manual startup needed
• Only stdio transport is supported (no TCP)
• Check logs in the API server output for errors

Development

To modify the system:

1. Add new tools: Edit stocks_server.py and add new @mcp.tool() functions
2. Update frontend: Modify frontend/src/App.tsx
3. Change API: Edit api_server.py for new endpoints
4. Update prompts: Modify the system message in api_server.py
5. Add logging: Use print() statements in tools for debugging

Project Structure

Stocks-Advisor-Agent-MCP/
├── api_server.py          # Flask API server
├── stocks_server.py       # MCP server with stock tools
├── client.py             # MCP client for testing
├── frontend/             # React frontend
│   ├── src/
│   ├── package.json
│   └── ...
└── README.md

License

This project is for educational and demonstration purposes.