Chat Web App for Azure AI Foundry Agent

Important

Blog post: Azure Developer CLI (azd): Debug hosted AI agents from your terminal

Code has been updated 4/28/2026 so that it works with hosted agents (v1.0 release of Microsoft Agent Framework).

A simple Flask web app with a chat window UI that calls an AI agent deployed to Azure AI Foundry. Designed for deployment to Azure App Service using azd.

Project Structure

simpleApp/
├── azure.yaml              # azd project configuration
├── infra/                  # Bicep infrastructure templates
│   ├── main.bicep
│   ├── main.parameters.json
│   └── modules/
│       ├── appservice.bicep
│       ├── ai-role-assignment.bicep
│       └── monitoring.bicep
└── src/                    # Flask application
    ├── app.py
    ├── requirements.txt
    ├── templates/
    │   └── index.html
    └── static/
        ├── style.css
        └── script.js

Prerequisites

• Azure Developer CLI (azd)
• Python 3.12+
• An Azure AI Foundry agent deployed and ready (you'll need the endpoint URL, agent name, and version)

Run Locally

cd src
pip install -r requirements.txt
set AZURE_AI_PROJECT_ENDPOINT="https://your-account.services.ai.azure.com/api/projects/your-project"
set AZURE_AI_AGENT_NAME="your-agent-name"
set AZURE_AI_AGENT_VERSION="1"
set AZURE_TENANT_ID="your-tenant-id"
python app.py

Open http://localhost:8000 in your browser.

Configure Your Foundry Agent

Before deploying, set the environment variables for your Foundry agent:

azd env set AZURE_AI_PROJECT_ENDPOINT "https://your-account.services.ai.azure.com/api/projects/your-project"
azd env set AZURE_AI_AGENT_NAME "your-agent-name"
azd env set AZURE_AI_AGENT_VERSION "3"
azd env set AZURE_TENANT_ID "your-tenant-id"

To enable automatic role assignment (recommended), also set your AI account details:

azd env set AI_ACCOUNT_RESOURCE_GROUP "your-ai-account-resource-group"
azd env set AI_ACCOUNT_NAME "your-ai-account-name"

Deploy to Azure

azd auth login
azd up

This will:

1. Provision an Azure App Service (Linux, Python 3.12) with Application Insights
2. Assign Azure AI Developer and Cognitive Services User roles to the App Service managed identity on your AI account (if AI_ACCOUNT_RESOURCE_GROUP and AI_ACCOUNT_NAME are set)
3. Deploy the Flask app
4. Output the URL of your deployed chat app

Authentication

The app uses DefaultAzureCredential to authenticate with Foundry. The App Service is created with a system-assigned managed identity. Role assignments are handled automatically via Bicep when the AI_ACCOUNT_RESOURCE_GROUP and AI_ACCOUNT_NAME environment variables are configured.