🚀 Azure AI Agent with MCP Tools

本案示範如何使用 Azure AI Agent 執行各種 AI 任務，包括基礎、中階、進階，以及與 MCP 整合的應用，並提供 Jupyter Notebook 及 Python 範例。

This repository demonstrates how to use Azure AI Agent for various AI tasks, including basic, intermediate, advanced, and MCP-integrated applications. It provides examples in both Jupyter Notebooks and Python scripts. Please see the user guide at the bottom.

1. 專案下載

下載本專案：

git clone https://github.com/ownway22/azure-ai-agent.git

進入專案資料夾：

cd azure-ai-agent

2. 環境建置

方法一：典型方式

1. 建立虛擬環境：

   python -m venv .venv

2. 啟用虛擬環境：
   • Windows PowerShell 指令：

     .venv\Scripts\Activate

   • CMD 指令：

     .venv\Scripts\activate.bat

   • macOS/Linux 指令：

     source .venv/bin/activate

3. 安裝所需套件：

   pip install -r requirements.txt

方法二：使用 uv sync

1. 快速建立與安裝虛擬環境：

   uv sync

2. 啟用虛擬環境：
   • Windows PowerShell 指令：

     .venv\Scripts\Activate

   • CMD 指令：

     .venv\Scripts\activate.bat

   • macOS/Linux 指令：

     source .venv/bin/activate

3. 修改環境變數

請到 Azure AI Foundry Project 複製您 AI project 中相關的環境變數，並將其填入 .env 檔案中。

PROJECT_ENDPOINT="https://*****.services.ai.azure.com/api/projects/*****"
API_KEY="86CBilkUxBKM3s8VYa7blR40bgIkDGhYnDys*****"
MODEL_DEPLOYMENT_NAME="gpt-4o"

AZURE_OPENAI_ENDPOINT="https://openai-ai-workshop.openai.azure.com/"
AZURE_OPENAI_API_KEY="AYj128C1av4Y9MLl01p5k6tIRY6pYRWOPjH*****"
AZURE_OPENAI_CHAT_DEPLOYMENT_NAME="gpt-4o"
AZURE_OPENAI_API_VERSION="2025-01-01-preview"
AZURE_BING_CONNECTION_ID="*****"

4. Azure CLI 登入 (az login)

在執行範例前，請先安裝並登入 Azure CLI：

1. 若尚未安裝 Azure CLI，請參考官方說明：如何安裝 Azure CLI
2. 登入 Azure 帳號：
   • Windows (PowerShell 或 CMD)：

     az login

   • macOS / Linux (Terminal)：

     az login

執行後會自動開啟瀏覽器進行帳號驗證，登入成功即可繼續後續步驟。

5. 開始執行範例

依序執行以下檔案：

1. 01_ai-agent-basic.ipynb
2. 02_ai-agent-intermediate.py
3. 03_ai-agent-advanced.py
4. 04_ai-agent-mcp-basic.ipynb
5. 05_ai-agent-mcp-intermediate.py
6. 06_ai-agent-mcp-advanced.py

執行 Jupyter Notebook 範例：

pip install notebook
jupyter notebook

然後在瀏覽器中開啟上述的 .ipynb 檔案。

執行 Python 範例：

python 02_ai-agent-intermediate.py
python 03_ai-agent-advanced.py
python 05_ai-agent-mcp-intermediate.py
python 06_ai-agent-mcp-advanced.py

6. 完成後的結果

當所有程式碼執行完成後，您可以在 Azure AI Foundry portal 的 agent 清單中看到如下畫面：

該畫面顯示已成功建立的 agent 清單，包含其名稱、ID、模型類型、建立時間等資訊。

---

The User Guide

This repository demonstrates how to use Azure AI Agent for various AI tasks, including basic, intermediate, advanced, and MCP-integrated applications. It provides examples in both Jupyter Notebooks and Python scripts.

1. Project Download

Download the repository:

git clone https://github.com/ownway22/azure-ai-agent.git

Navigate to the project folder:

cd azure-ai-agent

2. Environment Setup

Option 1: Typical Setup

1. Create a virtual environment:

   python -m venv .venv

2. Activate the virtual environment:
   • Windows PowerShell:

     .venv\Scripts\Activate

   • CMD:

     .venv\Scripts\activate.bat

   • macOS/Linux:

     source .venv/bin/activate

3. Install required packages:

   pip install -r requirements.txt

Option 2: Using uv sync

1. Quickly set up and install the virtual environment:

   uv sync

2. Activate the virtual environment:
   • Windows PowerShell:

     .venv\Scripts\Activate

   • CMD:

     .venv\Scripts\activate.bat

   • macOS/Linux:

     source .venv/bin/activate

3. Modify Environment Variables

Go to the Azure AI Foundry Project to copy the relevant environment variables for your AI project and add them to the .env file:

PROJECT_ENDPOINT="https://*****.services.ai.azure.com/api/projects/*****"
API_KEY="86CBilkUxBKM3s8VYa7blR40bgIkDGhYnDys*****"
MODEL_DEPLOYMENT_NAME="gpt-4o"

AZURE_OPENAI_ENDPOINT="https://openai-ai-workshop.openai.azure.com/"
AZURE_OPENAI_API_KEY="AYj128C1av4Y9MLl01p5k6tIRY6pYRWOPjH*****"
AZURE_OPENAI_CHAT_DEPLOYMENT_NAME="gpt-4o"
AZURE_OPENAI_API_VERSION="2025-01-01-preview"
AZURE_BING_CONNECTION_ID="*****"

4. Azure CLI Login (az login)

Before running the examples, install and log in to the Azure CLI:

1. If you haven't installed the Azure CLI yet, follow the official guide: Install the Azure CLI
2. Log in to your Azure account:
   • Windows (PowerShell or CMD):

     az login

   • macOS / Linux (Terminal):

     az login

This will automatically open a browser for account verification. Once logged in, you can proceed to the next steps.

5. Run Examples

Execute the following files in order:

1. 01_ai-agent-basic.ipynb
2. 02_ai-agent-intermediate.py
3. 03_ai-agent-advanced.py
4. 04_ai-agent-mcp-basic.ipynb
5. 05_ai-agent-mcp-intermediate.py
6. 06_ai-agent-mcp-advanced.py

Run Jupyter Notebook examples:

pip install notebook
jupyter notebook

Then open the .ipynb files in your browser.

Run Python script examples:

python 02_ai-agent-intermediate.py
python 03_ai-agent-advanced.py
python 05_ai-agent-mcp-intermediate.py
python 06_ai-agent-mcp-advanced.py

6. Results After Completion

After executing all the code, you can see the agent list in the Azure AI Foundry portal as shown below:

This screen displays the list of successfully created agents, including their names, IDs, model types, creation times, and other information.

---

Reference

01_ai-agent-basic.ipynb

1. Quickstart - Create a new Azure AI Foundry Agent Service project (2025/07)
2. Azure AI Projects client library for Python - version 1.0.0b12 (2025/06)
3. Azure AI Agents client library for Python - version 1.0.2 (2025/07)

02_ai-agent-intermediate.py

4. azure-sdk-for-python by dargilco (2025/06)

03_ai-agent-advanced.py

5. Build your code-first agent with Azure AI Foundry

04_ai-agent-mcp-basic.ipynb

6. Announcing Model Context Protocol Support (preview) in Azure AI Foundry Agent Service | Azure AI Foundry Blog (2025/06)
7. Code Samples for the Model Context Protocol Tool (Preview) - Azure AI Foundry | Microsoft Learn (2025/07)
8. Create a Remote MCP server for your GitHub repository

05_ai-agent-mcp-intermediate.py

9. QuickStart to MCP Tools Development with Azure AI Foundry SDK (2025/07)
10. AI-Foundry-Agent-MCP by Christian Coello (2025/07)

06_ai-agent-mcp-advanced.py

11. Build a MCP client using Azure AI Foundry and OpenAI Agents SDK (2025/04)
12. ai-foundry-agents-samples by Eitan Sela (2025/07)
13. Office PowerPoint MCP server
14. Excel Data Manager MCP server

More

15. AI Repo of the Week: MCP for Beginners | Microsoft Community Hub (2025/05)
16. 10 Microsoft MCP Servers to Accelerate Your Development Workflow - Microsoft for Developers (2025/07)