finance-llms/finmcp

本页由 AI 翻译，技术细节请以原文为准。查看原文

12 stars 5 forks 质量 78 趋势 16MIT

A unified Model Context Protocol (MCP) ecosystem for financial AI applications, providing deep research capabilities, trading integrations, and specialized AI model training for financial analysis.

概览

FinMCP is a comprehensive suite of interconnected financial AI tools designed to democratize institutional-grade financial analysis. Built on the Model Context Protocol (MCP) framework, it provides seamless integration between AI models and financial data sources, enabling sophisticated analysis, automated trading insights, and specialized AI model development for the financial domain. Demo Video Link :- https://drive.google.com/file/d/1sWKAB2K62oUG14YtS5YqEo72Unr9xfq3/view?usp=sharing - : AI-powered iterative research with web scraping and synthesis - : Real-time data from Indian (NSE/BSE) and global markets (Yahoo Finance) - : Direct connectivity with Upstox brokerage services - : Financial document analysis and LLM fine-tuning pipelines - : Standardized protocol for AI model connectivity - with pip - with npm - with submodule support - : NVIDIA, OpenAI, Firecrawl, Upstox (as needed) An AI-powered research assistant that performs iterative, comprehensive research on any...

README

以下为原文 README 节选，完整内容请查看原文。

FinMCP: Comprehensive Financial AI Ecosystem

A unified Model Context Protocol (MCP) ecosystem for financial AI applications, providing deep research capabilities, trading integrations, and specialized AI model training for financial analysis.

🎯 Overview

Demo Video Link :- https://drive.google.com/file/d/1sWKAB2K62oUG14YtS5YqEo72Unr9xfq3/view?usp=sharing

Key Innovations

🔬 Deep Research Engine: AI-powered iterative research with web scraping and synthesis
📈 Multi-Market Integration: Real-time data from Indian (NSE/BSE) and global markets (Yahoo Finance)
🏦 Trading Platform Integration: Direct connectivity with Upstox brokerage services
🧠 Specialized AI Training: Financial document analysis and LLM fine-tuning pipelines
🔌 MCP Architecture: Standardized protocol for AI model connectivity

🏗️ Architecture

graph TB
    subgraph "FinMCP Ecosystem"
        subgraph "Core MCP Framework"
            MCP[MCP Protocol Layer]
            CONFIG[mcp.json Configuration]
        end
        
        subgraph "Research & Analysis"
            DR[Deep Research Engine]
            DP[Digital Persona AI]
        end
        
        subgraph "Market Data Sources"
            YF[Yahoo Finance Global]
            IM[Indian Markets NSE/BSE]
        end
        
        subgraph "Trading Integration"
            UP[Upstox Trading Platform]
        end
        
        subgraph "AI Models"
            LLM[LLMs via NVIDIA/OpenAI]
            FT[Fine-tuned Financial Models]
        end
    end
    
    CONFIG --> DR
    CONFIG --> DP
    CONFIG --> YF
    CONFIG --> IM
    CONFIG --> UP
    
    DR --> LLM
    DP --> FT
    YF --> LLM
    IM --> LLM
    UP --> LLM
    
    style MCP fill:#e1f5fe
    style DR fill:#f3e5f5
    style DP fill:#e8f5e8
    style YF fill:#fff3e0
    style IM fill:#fce4ec
    style UP fill:#e0f2f1

🚀 Quick Start

Prerequisites

Python 3.8+ with pip
Node.js 16+ with npm
Git with submodule support
API Keys: NVIDIA, OpenAI, Firecrawl, Upstox (as needed)

Installation

Clone the repository with submodules:

git clone --recursive https://github.com/Finance-LLMs/FinMCP.git
cd FinMCP

Update submodules to latest:

在 GitHub 查看完整 README

安装

npx mcp-remote http://localhost:8787/sse

新手？先看 mcpskill.org 教程 →

配置

{
  "mcpServers": {
    "mcp-server-upstox-api": {
      "command": "npx",
      "args": ["mcp-remote", "http://localhost:8787/sse"]
    },
    "Stocks": {
      "command": "uv",
      "args": ["run", "--with", "mcp[cli]", "--with", "yfinance", "mcp", "run", "Yahoo-Finance/main.py"]
    },
    "IndianMarkets": {
      "command": "uv", 
      "args": ["run", "--with", "mcp[cli]", "--with", "bse", "--with", "nsepython", "mcp", "run", "Indian-Markets/indian_markets.py"]
    },
    "deep-research-python": {
      "command": "python3",
      "args": ["Deep-Research/mcp_deep_research.py"],
      "env": {
        "PYTHONPATH": "Deep-Research/"
      }
    }
  }
}