AI Agent Using MCP Use

带有MCP服务器的AI代理

一个交互式AI代理，利用模型上下文协议（MCP）服务器执行各种任务，包括网页浏览、Airbnb搜索和使用Blender创建3D内容。具有用户友好的控制台界面，可进行实时交互。

特性

• 🌐 网页浏览：使用Playwright进行自动网络导航和搜索
• 🏠 Airbnb集成：搜索和检索Airbnb房源详细信息
• 🎨 3D创建：使用Blender生成和操纵3D内容
• 🤖 智能代理：由Groq的快速推理LLMs提供支持
• 💬 交互式控制台：与AI代理的实时聊天界面
• 🔧 可扩展：易于添加更多MCP服务器和功能

演示

============================================================
🤖 AI Agent with MCP Servers
============================================================
Available capabilities:
  🌐 Web browsing and search (Playwright)
  🏠 Airbnb search and listings
  🎨 3D content creation (Blender)
============================================================
Commands:
  - Type your query and press Enter
  - Type 'quit', 'exit', or 'q' to exit
  - Type 'help' for example queries
============================================================

💬 Enter your query: Find the best restaurants in Tokyo

🔄 Processing: Find the best restaurants in Tokyo
--------------------------------------------------

✅ Result:
--------------------------------------------------
Based on my search, here are the top restaurants in Tokyo...
--------------------------------------------------

💬 Enter your query: 

先决条件

• Python 3.8或更高版本
• Node.js和npm（用于MCP服务器）
• UV包管理器
• Groq API 密钥

安装

1. 克隆存储库：

   git clone https://github.com/yourusername/ai-agent-mcp.git
   cd ai-agent-mcp

1. 安装Python依赖项：

   uv sync

1. 安装MCP服务器：

   # Playwright MCP server (for web browsing)
   npx @playwright/mcp@latest

   # Airbnb MCP server
   npx -y @openbnb/mcp-server-airbnb

   # Blender MCP server
   uvx blender-mcp

1. 设置环境变量：

创建一个 .env 项目根目录中的文件：

   GROQ_API_KEY=your_groq_api_key_here

快速开始

1. 运行交互式代理：

   uv run main.py

1. 开始与代理聊天：

- 键入您的查询，然后按Enter键 - 使用 help 查看示例查询 - 使用 quit, exit，或 q 退出

使用示例

Web搜索查询

💬 Enter your query: Search for the latest news about artificial intelligence
💬 Enter your query: Find the weather forecast for New York City
💬 Enter your query: Navigate to GitHub and search for Python projects

Airbnb查询

💬 Enter your query: Find Airbnb listings in Paris for 2 guests under $100 per night
💬 Enter your query: Search for beachfront accommodations in Miami
💬 Enter your query: Show me luxury apartments in Tokyo for next month

三维创建查询

💬 Enter your query: Create a 3D cube in Blender and take a screenshot
💬 Enter your query: Generate a simple house model in Blender
💬 Enter your query: Add a texture to a sphere in Blender

控制台命令

配置

MCP服务器配置在 mcp_server_config.json。您可以修改此文件以添加、删除或配置服务器：

{
  "mcpServers": {
    "playwright": {
      "command": "npx",
      "args": ["@playwright/mcp@latest"],
      "env": {
        "DISPLAY": ":1"
      }
    },
    "airbnb": {
      "command": "npx",
      "args": ["-y", "@openbnb/mcp-server-airbnb"]
    },
    "blender": {
      "command": "uvx", 
      "args": ["blender-mcp"]
    }
  }
}

自定义LLM模型

您可以在中更改AI模型 main.py:

# Options for different use cases:
llm = ChatGroq(model="llama3-8b-8192")        # Fast, good for most tasks
llm = ChatGroq(model="llama3-70b-8192")       # More capable, better reasoning
llm = ChatGroq(model="llama-3.1-8b-instant")  # Large context window (128K)