FinSight: AI-Powered Financial Analyst 📈

FinSight is an autonomous financial analysis agent that integrates directly into your IDE via the Model Context Protocol (MCP). Built with CrewAI and Deepseek-R1, it analyzes stock market data, fetches historical prices, and provides intelligent financial insights right where you work.

Instead of switching back and forth between financial websites and your code, you can just ask your IDE to analyze stock performance, compare companies, and interpret market trends.

🚀 Features

Local AI Execution: Uses Deepseek-R1 via Ollama to keep your analysis private and local.
Multi-Agent Orchestration: Powered by CrewAI to break down complex financial queries into specialized agent tasks.
IDE Integration: Runs natively as an MCP server in Cursor or any compatible IDE.
Real-Time Data: Fetches stock prices, volume, and trends using yfinance.

🛠️ Getting Started

1. Prerequisites

Python 3.11+
Docker (Required by CrewAI for secure code execution)
Ollama (For running Deepseek-R1 locally)

2. Install & Run Ollama

You'll need Ollama running in the background with the Deepseek-R1 model:

ollama run deepseek-r1

3. Install Dependencies

This project uses uv for fast dependency management.

uv sync --python 3.11

4. Connect to Your IDE (Cursor)

To use FinSight directly from your IDE, you need to add it as an MCP server.

Open your IDE settings and navigate to MCP.
Add a new global MCP server with the following configuration:

{
    "mcpServers": {
        "financial-analyst": {
            "command": "uv",
            "args": [
                "--directory",
                "C:/absolute/path/to/FinSight",
                "run",
                "server.py"
            ]
        }
    }
}

Restart or reload the MCP server in your IDE.

💡 How to Use It

Once connected, you can chat with your IDE and ask financial questions like:

"Show me Tesla's stock performance over the last 3 months."
"Compare Apple and Microsoft stocks for the past year."
"Analyze the trading volume of Amazon stock for the last month."

The AI will spin up the necessary CrewAI agents, fetch the data, and deliver a comprehensive analysis!