Implementation:Microsoft Agent framework Pip Install Agent Framework

Property	Value
Implementation Name	Pip Install Agent Framework
SDK	Microsoft Agent Framework
Repository	agent-framework
Source File	python/packages/core/pyproject.toml
Line Range	L1-132
Type	External Tool Doc
Domains	AI_Infrastructure, DevOps

Overview

The Pip Install Agent Framework implementation covers the installation of the agent-framework-core package via pip and the configuration of environment variables required for LLM provider authentication. The package specification is defined in python/packages/core/pyproject.toml and targets Python >= 3.10. Installation supports optional extras for Azure integration, while companion packages provide declarative agent definitions and multi-agent orchestrations.

Code Reference

Source Location

Property	Value
File	python/packages/core/pyproject.toml
Lines	1-132
Package Name	agent-framework-core
Python Requirement	>= 3.10

Key Dependencies

Dependency	Version Constraint	Purpose
pydantic	>= 2	Data validation and schema enforcement for agent configuration, tool parameters, and model options.
openai	>= 1.99.0	OpenAI Python SDK providing the Responses API client.
azure-identity	>= 1	Azure credential management for authenticating with Azure OpenAI endpoints (via [azure] extra).
opentelemetry-api	>= 1.39.0	Distributed tracing and observability hooks for agent execution pipelines.

I/O Contract

Inputs

Input	Type	Required	Description
pip install command	Shell command	Yes	The pip install invocation specifying the package name and optional extras.
OPENAI_API_KEY	Environment variable	Conditional	API key for OpenAI. Required when using OpenAIResponsesClient.
OPENAI_RESPONSES_MODEL_ID	Environment variable	No	Model identifier for the Responses API (e.g., gpt-4o). Falls back to a default if unset.
AZURE_OPENAI_API_KEY	Environment variable	Conditional	API key for Azure OpenAI. Required when using AzureOpenAIResponsesClient with key-based auth.
AZURE_OPENAI_ENDPOINT	Environment variable	Conditional	Azure OpenAI resource endpoint URL. Required for Azure deployments.
AZURE_OPENAI_RESPONSES_DEPLOYMENT_NAME	Environment variable	Conditional	Deployment name within the Azure OpenAI resource. Required for Azure deployments.

Output

Output	Type	Description
Installed Python packages	Filesystem artifacts	The agent-framework-core package and its dependencies installed into the active Python environment.
Configured runtime environment	Environment state	Environment variables set in the shell session, enabling OpenAIResponsesClient or AzureOpenAIResponsesClient to authenticate without explicit credential parameters.

Installation Commands

Core Package

# Install the core package with base dependencies
pip install agent-framework-core

With Optional Extras

# Install with all optional extras
pip install agent-framework-core[all]

# Install with Azure extras only
pip install agent-framework-core[azure]

Companion Packages

# Declarative agent definitions (YAML/JSON-based agent construction)
pip install agent-framework-declarative

# Multi-agent orchestration patterns (group chat, handoff, pipelines)
pip install agent-framework-orchestrations

Environment Variable Configuration

OpenAI

export OPENAI_API_KEY="sk-..."
export OPENAI_RESPONSES_MODEL_ID="gpt-4o"

Azure OpenAI

export AZURE_OPENAI_API_KEY="..."
export AZURE_OPENAI_ENDPOINT="https://your-resource.openai.azure.com/"
export AZURE_OPENAI_RESPONSES_DEPLOYMENT_NAME="gpt-4o"

Usage Examples

Verify Installation with OpenAI

from agent_framework import Agent
from agent_framework.openai import OpenAIResponsesClient

# Client reads OPENAI_API_KEY from environment automatically
client = OpenAIResponsesClient()

agent = Agent(
    client=client,
    instructions="You are a helpful assistant.",
    name="install-test",
)

# Run a simple query to verify end-to-end connectivity
response = await agent.run("Hello, are you working?")
print(response)

Verify Installation with Azure OpenAI

from agent_framework import Agent
from agent_framework.azure import AzureOpenAIResponsesClient

# Client reads AZURE_OPENAI_* variables from environment automatically
client = AzureOpenAIResponsesClient()

agent = Agent(
    client=client,
    instructions="You are a helpful assistant.",
    name="azure-install-test",
)

response = await agent.run("Hello from Azure!")
print(response)

Related Pages

• Principle:Microsoft_Agent_framework_Framework_Installation
• Environment:Microsoft_Agent_framework_Python_3_10_Runtime
• Environment:Microsoft_Agent_framework_Core_Package_Dependencies