Implementation:Microsoft Agent framework AzureOpenAIResponsesClient Init

Knowledge Sources	Microsoft Agent Framework Azure OpenAI Service
Domains	AI_Infrastructure, Azure, Agent_Architecture
Last Updated	2026-02-11 17:00 GMT

Overview

Concrete tool for configuring an Azure OpenAI Responses API client with Azure AD or API key authentication, provided by the Microsoft Agent Framework.

Description

The AzureOpenAIResponsesClient initializes an authenticated connection to an Azure OpenAI deployment. It supports direct endpoint connection with API key or Azure AD credentials, as well as creation through an Azure AI Foundry project endpoint. The client extends mixin layers for middleware, function invocation, and telemetry, implementing the SupportsChatGetResponse protocol.

Usage

Import this class when deploying agents against Azure OpenAI. Provide either api_key and endpoint for key-based auth, or credential for Azure AD authentication. Use project_endpoint or project_client for Foundry-based creation requiring the azure-ai-projects package.

Code Reference

Source Location

Repository: agent-framework
File: python/packages/core/agent_framework/azure/_responses_client.py
Lines: L55-245

Signature

class AzureOpenAIResponsesClient(
    AzureOpenAIConfigMixin,
    ChatMiddlewareLayer[AzureOpenAIResponsesOptionsT],
    FunctionInvocationLayer[AzureOpenAIResponsesOptionsT],
    ChatTelemetryLayer[AzureOpenAIResponsesOptionsT],
    RawOpenAIResponsesClient[AzureOpenAIResponsesOptionsT],
    Generic[AzureOpenAIResponsesOptionsT],
):
    def __init__(
        self,
        *,
        api_key: str | None = None,
        deployment_name: str | None = None,
        endpoint: str | None = None,
        base_url: str | None = None,
        api_version: str | None = None,
        ad_token: str | None = None,
        ad_token_provider: AsyncAzureADTokenProvider | None = None,
        token_endpoint: str | None = None,
        credential: TokenCredential | None = None,
        default_headers: Mapping[str, str] | None = None,
        async_client: AsyncOpenAI | None = None,
        project_client: Any | None = None,
        project_endpoint: str | None = None,
        env_file_path: str | None = None,
        env_file_encoding: str | None = None,
        instruction_role: str | None = None,
        middleware: Sequence[MiddlewareTypes] | None = None,
        function_invocation_configuration: FunctionInvocationConfiguration | None = None,
        **kwargs: Any,
    ) -> None:

Import

from agent_framework.azure import AzureOpenAIResponsesClient

I/O Contract

Inputs

Name	Type	Required	Description
api_key	str	No	Azure OpenAI API key. Also via env var AZURE_OPENAI_API_KEY.
deployment_name	str	No	Azure deployment name. Also via env var AZURE_OPENAI_RESPONSES_DEPLOYMENT_NAME.
endpoint	str	No	Azure OpenAI endpoint URL. Also via env var AZURE_OPENAI_ENDPOINT.
base_url	str	No	Full base URL (must end with /openai/v1/).
api_version	str	No	API version (must be "preview"). Also via env var AZURE_OPENAI_API_VERSION.
credential	TokenCredential	No	Azure credential for Azure AD authentication.
project_client	AIProjectClient	No	Pre-configured Foundry project client.
project_endpoint	str	No	Azure AI Foundry project endpoint URL.
middleware	Sequence	No	Middleware pipeline for request/response interception.

Outputs

Name	Type	Description
instance	AzureOpenAIResponsesClient	Configured client implementing SupportsChatGetResponse protocol.

Usage Examples

With Azure CLI Credential

from agent_framework.azure import AzureOpenAIResponsesClient
from azure.identity import AzureCliCredential

client = AzureOpenAIResponsesClient(
    credential=AzureCliCredential(),
)

With API Key

from agent_framework.azure import AzureOpenAIResponsesClient

client = AzureOpenAIResponsesClient(
    api_key="your-api-key",
    endpoint="https://your-resource.openai.azure.com/",
    deployment_name="gpt-4o",
)

With Foundry Project Endpoint

from agent_framework.azure import AzureOpenAIResponsesClient
from azure.identity import DefaultAzureCredential

client = AzureOpenAIResponsesClient(
    project_endpoint="https://your-project.services.ai.azure.com/api/projects/your-project",
    credential=DefaultAzureCredential(),
)

Related Pages

Implements Principle

Page Connections

Double-click a node to navigate. Hold to expand connections.

Principle

Implementation

Heuristic

Environment