agent-framework/dotnet/samples/02-agents/AgentsWithFoundry/Agent_Step01_Basics

Creating and Running a Basic Agent with the Responses API

This sample demonstrates how to create and run a basic AI agent using the ChatClientAgent, which uses the Microsoft Foundry Responses API directly without creating server-side agent definitions.

What this sample demonstrates

• Creating a ChatClientAgent with instructions and a model
• Running a simple single-turn conversation
• No server-side agent creation or cleanup required

Prerequisites

Before you begin, ensure you have the following prerequisites:

• .NET 10 SDK or later
• Microsoft Foundry service endpoint and deployment configured
• Azure CLI installed and authenticated (for Azure credential authentication)

Note: This demo uses Azure CLI credentials for authentication. Make sure you're logged in with az login and have access to the Microsoft Foundry resource. For more information, see the Azure CLI documentation.

Set the following environment variables:

$env:AZURE_AI_PROJECT_ENDPOINT="https://your-foundry-service.services.ai.azure.com/api/projects/your-foundry-project"
$env:AZURE_AI_MODEL_DEPLOYMENT_NAME="gpt-5.4-mini"

Run the sample

Navigate to the AgentsWithFoundry sample directory and run:

cd dotnet/samples/02-agents/AgentsWithFoundry
dotnet run --project .\Agent_Step01_Basics

Alternative: Composable approach

You can also create the same agent by composing the underlying IChatClient directly. This gives you full control over the chat client pipeline:

using Azure.AI.Projects;
using Azure.Identity;
using Microsoft.Agents.AI;
using Microsoft.Extensions.AI;

AIProjectClient aiProjectClient = new(new Uri(endpoint), new DefaultAzureCredential());

AIAgent agent = new ChatClientAgent(
    chatClient: aiProjectClient.GetProjectOpenAIClient().GetProjectResponsesClient().AsIChatClient(deploymentName),
    instructions: "You are good at telling jokes.",
    name: "JokerAgent");

This approach is useful when you need to customize the chat client pipeline or swap providers (e.g., Anthropic, OpenAI) while keeping the same agent code.