Code Interpreter tool for Microsoft Foundry agents

In this article, you create an agent that uses Code Interpreter to execute Python code in a sandboxed environment. You learn how to enable agents to solve mathematical problems, analyze data from CSV files, and generate visualizations like charts and graphs.

Code Interpreter enables agents to write and run Python code in a sandboxed execution environment. By enabling Code Interpreter, your agent can run code iteratively to solve more challenging code, math, and data analysis problems or create graphs and charts. When your agent writes code that doesn't run, it can modify and run different code until the code execution succeeds.

Usage support

Prerequisites

• Basic or standard agent environment. See agent environment setup for details.
• Latest prerelease SDK package installed. See the quickstart for installation steps.
• Azure AI model deployment configured in your project.
• For file operations: CSV or other supported files to upload for analysis.

Code samples

import os
import httpx
from dotenv import load_dotenv
from azure.identity import DefaultAzureCredential
from azure.ai.projects import AIProjectClient
from azure.ai.projects.models import PromptAgentDefinition, CodeInterpreterTool, CodeInterpreterToolAuto

load_dotenv()

# Load the CSV file to be processed
asset_file_path = os.path.abspath(
    os.path.join(os.path.dirname(__file__), "../assets/synthetic_500_quarterly_results.csv")
)

project_client = AIProjectClient(
    endpoint=os.environ["AZURE_AI_PROJECT_ENDPOINT"],
    credential=DefaultAzureCredential(),
)

with project_client:
    openai_client = project_client.get_openai_client()

    # Upload the CSV file for the code interpreter to use
    file = openai_client.files.create(purpose="assistants", file=open(asset_file_path, "rb"))
    print(f"File uploaded (id: {file.id})")

    # Create agent with code interpreter tool
    agent = project_client.agents.create_version(
        agent_name="MyAgent",
        definition=PromptAgentDefinition(
            model=os.environ["AZURE_AI_MODEL_DEPLOYMENT_NAME"],
            instructions="You are a helpful assistant.",
            tools=[CodeInterpreterTool(container=CodeInterpreterToolAuto(file_ids=[file.id]))],
        ),
        description="Code interpreter agent for data analysis and visualization.",
    )
    print(f"Agent created (id: {agent.id}, name: {agent.name}, version: {agent.version})")

    # Create a conversation for the agent interaction
    conversation = openai_client.conversations.create()
    print(f"Created conversation (id: {conversation.id})")

    # Send request to create a chart and generate a file
    response = openai_client.responses.create(
        conversation=conversation.id,
        input="Could you please create bar chart in TRANSPORTATION sector for the operating profit from the uploaded csv file and provide file to me?",
        extra_body={"agent": {"name": agent.name, "type": "agent_reference"}},
    )
    print(f"Response completed (id: {response.id})")

    # Extract file information from response annotations
    file_id = ""
    filename = ""
    container_id = ""

    # Get the last message which should contain file citations
    last_message = response.output[-1]  # ResponseOutputMessage
    if last_message.type == "message":
        # Get the last content item (contains the file annotations)
        text_content = last_message.content[-1]  # ResponseOutputText
        if text_content.type == "output_text":
            # Get the last annotation (most recent file)
            if text_content.annotations:
                file_citation = text_content.annotations[-1]  # AnnotationContainerFileCitation
                if file_citation.type == "container_file_citation":
                    file_id = file_citation.file_id
                    filename = file_citation.filename
                    container_id = file_citation.container_id
                    print(f"Found generated file: {filename} (ID: {file_id})")

    # Download the generated file if available
    if file_id and filename:
        file_content = openai_client.containers.files.content.retrieve(file_id=file_id, container_id=container_id)
        with open(filename, "wb") as f:
            f.write(file_content.read())
            print(f"File {filename} downloaded successfully.")
        print(f"File ready for download: {filename}")
    else:
        print("No file generated in response")
    #uncomment these lines if you want to delete your agent
    #print("\nCleaning up...")
    #project_client.agents.delete_version(agent_name=agent.name, agent_version=agent.version)
    #print("Agent deleted")

File uploaded (id: file-xxxxxxxxxxxxxxxxxxxx)
Agent created (id: agent-xxxxxxxxxxxxxxxxxxxx, name: MyAgent, version: 1)
Created conversation (id: conv-xxxxxxxxxxxxxxxxxxxx)
Response completed (id: resp-xxxxxxxxxxxxxxxxxxxx)
Found generated file: transportation_operating_profit_bar_chart.png (ID: file-xxxxxxxxxxxxxxxxxxxx)
File transportation_operating_profit_bar_chart.png downloaded successfully.
File ready for download: transportation_operating_profit_bar_chart.png

// Create project client and read the environment variables, which will be used in the next steps.
var projectEndpoint = System.Environment.GetEnvironmentVariable("PROJECT_ENDPOINT");
var modelDeploymentName = System.Environment.GetEnvironmentVariable("MODEL_DEPLOYMENT_NAME");
AIProjectClient projectClient = new(endpoint: new Uri(projectEndpoint), tokenProvider: new DefaultAzureCredential());

// Create Agent, capable to use Code Interpreter to answer questions.
PromptAgentDefinition agentDefinition = new(model: modelDeploymentName)
{
    Instructions = "You are a helpful agent that can help fetch data from files you know about.",
    Tools = {
        ResponseTool.CreateCodeInterpreterTool(
            new CodeInterpreterToolContainer(
                CodeInterpreterToolContainerConfiguration.CreateAutomaticContainerConfiguration(
                    fileIds: []
                )
            )
        ),
    }
};
AgentVersion agentVersion = projectClient.Agents.CreateAgentVersion(
    agentName: "myAgent",
    options: new(agentDefinition));

// Ask the agent a question, which requires running python code in the container.
AgentReference agentReference = new(name: agentVersion.Name, version: agentVersion.Version);
ProjectResponsesClient responseClient = projectClient.OpenAI.GetProjectResponsesClientForAgent(agentReference);

ResponseResult response = responseClient.CreateResponse("I need to solve the equation sin(x) + x^2 = 42");

// Write out the output of a response, raise the exception if the request was not successful.
Assert.That(response.Status, Is.EqualTo(ResponseStatus.Completed));
Console.WriteLine(response.GetOutputText());

// Clean up resources by deleting conversations and the Agent.
projectClient.Agents.DeleteAgentVersion(agentName: agentVersion.Name, agentVersion: agentVersion.Version);

Response completed (id: resp-xxxxxxxxxxxxxxxxxxxx)
The solution to the equation sin(x) + x^2 = 42 is approximately x = 6.324555320336759

Regional restrictions

The code interpreter tool for the Foundry projects (new) API isn't available in the following regions:

• Canada Central
• Central US
• Japan East
• South Central US
• Southeast Asia
• Spain Central

Supported file types