AI导哈

AI智能体与MCP集成集合

本仓库提供一系列AI智能体实现示例，涵盖金融数据分析、多智能体协作和MCP服务器，帮助开发者学习并构建智能体应用。

  • by rgull
  • Csound
  • 本地部署
  • 更新于
在 GitHub 查看 rgull/MCP-Agent

使用说明

🤖 AI Agents with MCP

A comprehensive collection of AI agent implementations showcasing various frameworks, models, and use cases. This repository contains practical examples of building intelligent agents using Phidata and MCP (Model Context Protocol).

📚 Table of Contents

🎯 Overview

This repository serves as a learning resource for building AI agents with different capabilities:

  • Finance Agents: Stock market analysis, company research, and financial data retrieval
  • Multi-Agent Teams: Collaborative agents working together to solve complex tasks
  • MCP Servers: Model Context Protocol servers for Claude Desktop integration

🚀 Projects

1. Phidata Finance Agent (1_phidata_finance_agent/)

A collection of finance-focused AI agents demonstrating various capabilities:

📄 1_simple_groq_agent.py

A basic agent setup using Groq's Llama 3.3 70B model. Perfect starting point for understanding agent fundamentals.

Features:

  • Simple agent initialization
  • Groq model integration
  • Basic conversation capabilities

📄 2_finance_agent_llama.py

An advanced finance agent that can analyze stocks, compare companies, and retrieve financial data.

Features:

  • Real-time stock price retrieval
  • Analyst recommendations analysis
  • Company fundamentals comparison
  • Custom company symbol mapping
  • Table-based data visualization

Example Use Cases:

  • Compare analyst recommendations for multiple stocks
  • Analyze company fundamentals
  • Get real-time stock prices

📄 3_agent_teams_openai.py

A multi-agent system where specialized agents collaborate to provide comprehensive answers.

Features:

  • Web Agent: Searches the internet for latest news and information
  • Finance Agent: Retrieves financial data and stock information
  • Team Coordinator: Orchestrates both agents to provide complete answers
  • OpenAI GPT-4o integration
  • Source attribution

Example Use Cases:

  • Get latest news + financial analysis for a stock
  • Research companies with both web and financial data
  • Comprehensive market analysis

2. MCP Leave Management (2_mcp_leave_management/)

A Model Context Protocol (MCP) server for employee leave management that integrates with Claude Desktop.

Features:

  • Check leave balance
  • Apply for leave with specific dates
  • View leave history
  • Personalized greetings
  • In-memory database with mock employee data

Perfect for:

  • Learning MCP server development
  • Building HR management tools
  • Claude Desktop integration examples

✨ Features

  • 🔌 Multiple Model Support: Groq, OpenAI, and more
  • 🛠️ Rich Tool Integration: YFinance, DuckDuckGo, custom tools
  • 👥 Multi-Agent Collaboration: Teams of specialized agents
  • 📊 Data Visualization: Table-based output for better readability
  • 🔍 Web Search Capabilities: Real-time information retrieval
  • 💼 Finance Tools: Stock prices, analyst recommendations, fundamentals
  • 🖥️ MCP Integration: Claude Desktop server examples

📋 Prerequisites

Before you begin, ensure you have the following installed:

  • Python 3.10 or higher
  • pip or uv package manager
  • API keys for:
    • Groq API (for Groq models)
    • OpenAI API (for GPT models)

🛠️ Getting Started

1. Clone the Repository

git clone <repository-url>
cd ai-agents

2. Install Dependencies

For Phidata Agents

# Install Phidata and dependencies
pip install phidata yfinance duckduckgo-search python-dotenv

# Or using uv
uv pip install phidata yfinance duckduckgo-search python-dotenv

For MCP Leave Management

cd 2_mcp_leave_management
uv add "mcp[cli]>=1.6.0"

3. Set Up Environment Variables

Create a .env file in the project root:

GROQ_API_KEY=your_groq_api_key_here
OPENAI_API_KEY=your_openai_api_key_here

4. Run the Agents

Simple Groq Agent

python 1_phidata_finance_agent/1_simple_groq_agent.py

Finance Agent

python 1_phidata_finance_agent/2_finance_agent_llama.py

Agent Teams

python 1_phidata_finance_agent/3_agent_teams_openai.py

MCP Server

cd 2_mcp_leave_management
uv run mcp install main.py
# Restart Claude Desktop to see the tools

📁 Project Structure

ai-agents/
├── 1_phidata_finance_agent/
│   ├── 1_simple_groq_agent.py      # Basic agent example
│   ├── 2_finance_agent_llama.py    # Finance agent with tools
│   └── 3_agent_teams_openai.py     # Multi-agent team
├── 2_mcp_leave_management/
│   ├── main.py                     # MCP server implementation
│   ├── pyproject.toml              # Project dependencies
│   ├── uv.lock                     # Lock file
│   └── README.md                   # MCP-specific documentation
└── README.md                       # This file

💡 Usage Examples

Finance Agent Example

agent.print_response(
    "Summarize and compare analyst recommendations and fundamentals for TSLA and MSFT. Show in tables.",
    stream=True
)

Multi-Agent Team Example

agent_team.print_response(
    "Summarize analyst recommendations and share the latest news for NVDA",
    stream=True
)

MCP Server Usage

Once installed in Claude Desktop, you can use commands like:

  • "Check leave balance for employee E001"
  • "Apply for leave on 2025-04-17 and 2025-05-01 for E001"
  • "Show leave history for E002"

🛠️ Technologies Used

  • Phidata: Framework for building AI agents
  • Groq: Fast inference API for LLMs
  • OpenAI: GPT-4o and other models
  • YFinance: Financial data retrieval
  • DuckDuckGo: Web search capabilities
  • MCP: Model Context Protocol for Claude Desktop
  • FastMCP: Fast MCP server framework

🤝 Contributing

Contributions are welcome! Feel free to:

  1. Fork the repository
  2. Create a feature branch
  3. Make your changes
  4. Submit a pull request

📝 License

This project is for educational purposes. Please check individual project licenses.

🙏 Acknowledgments

  • Phidata team for the excellent agent framework
  • Codebasics Inc. / LearnerX Pvt Ltd for MCP examples

Happy Building! 🚀

For questions or issues, please open an issue on GitHub.

相关 MCP