Creating Your First Agent — Building AI Agents

[ ABORT TO HUD ]
SEQ. 1
SEQ. 2
SEQ. 3

Creating Your First Agent

🤖 Building AI Agents 10 min 80 BASE XP⌨ HANDS-ON LAB

Building an Agent in 5 Minutes

Via the Foundry Portal

  1. Navigate to Agents in your project sidebar
  2. Click + New Agent
  3. Select a deployed model (e.g., gpt-4o)
  4. Write system instructions defining the agent's role
  5. Add tools (code interpreter, file search, custom functions)
  6. Test in the Agent Playground

Via the SDK (Python)

from azure.ai.projects import AIProjectClient
from azure.ai.projects.models import PromptAgentDefinition, CodeInterpreterTool, MCPTool
from azure.identity import DefaultAzureCredential

with AIProjectClient(
    endpoint="https://<foundry-name>.services.ai.azure.com/api/projects/<project>",
    credential=DefaultAzureCredential()
) as project:
    # 1. Create a versioned agent in Foundry Agent Service (v2 SDK)
    agent = project.agents.create_version(
        agent_name="ResearchAssistant",
        definition=PromptAgentDefinition(
            model="gpt-6.1-sol",
            instructions="You are a research assistant. Analyze data and cite sources clearly.",
            tools=[
                CodeInterpreterTool(),
                MCPTool(server_label="docs-mcp", server_url="https://learn.microsoft.com/api/mcp")
            ],
        ),
    )

    # 2. Run a stateful Conversation via the OpenAI Responses API
    with project.get_openai_client() as openai_client:
        conversation = openai_client.conversations.create()
        response = openai_client.responses.create(
            conversation=conversation.id,
            extra_body={"agent": {"name": agent.name, "type": "agent_reference"}},
            input="Analyze the latest trends in renewable energy",
        )
        print(response.output_text)
🎯 Pro Tip: Write agent instructions as if briefing a new employee. Be specific about what the agent should and should NOT do, what tone to use, and how to handle edge cases.
⌨ HANDS-ON LABCreate Foundry Agent with Code Interpreter Tool
⭐ +200 XP

Use Python azure-ai-projects SDK to declare a new autonomous agent equipped with sandboxed Python code interpreter.

1Execute Python code declaring an agent with instructions and the code_interpreter tool definition.
lab-sandbox — simulated environment
INFINITY LAB SANDBOX v2.6 — simulated shell
Type the command for the current objective. Helpers: "hint", "solution", "clear".
$
OBJECTIVE 1 / 1 — type "hint" if stuck
FOUNDRY VERIFICATION
QUERY 1 // 1
What is the recommended approach for writing agent instructions?
Keep them as short as possible
Write them like briefing a new employee - specific about role, boundaries, and edge cases
Copy from documentation
Leave them empty and rely on the model