feat: Agent framework - Upgrading to new Foundry Agents changes

[Abdul-Microsoft]

Purpose

• This pull request introduces significant updates to the agent orchestration and Azure AI integration, focusing on upgrading dependencies, refactoring agent client usage, improving logging configuration, and ensuring safer team configuration handling. The main themes are dependency upgrades, Azure AI client migration, code cleanup, and improved robustness in agent/team management.

Dependency and SDK Upgrades:

• Upgraded several core dependencies in pyproject.toml, including azure-ai-projects (to 2.x), azure-monitor-opentelemetry, openai, and all opentelemetry packages, and replaced agent-framework with agent-framework-azure-ai and agent-framework-core for improved Azure AI support. [1] [2]

Azure AI Client Migration and Refactoring:

• Refactored agent client usage throughout the codebase, replacing AzureAIAgentClient with the new AzureAIClient from agent_framework_azure_ai, and updated all related class methods and instantiation patterns accordingly. This includes removing legacy agent ID resolution and database agent reuse logic, as the new SDK manages agent reuse by name. [1] [2] [3] [4] [5]
• Updated the foundry agent and related workflow logic to use the new Azure AI client and models, aligning with the upgraded SDK. [1] [2]

Logging and Configuration Improvements:

• Moved logging configuration to the very top of app.py to ensure logging levels are set before any logging calls, and added explicit log level setting for Azure HTTP logging policy. [1] [2] [3]

Agent/Team Management Robustness:

• In the RAI agent creation utility, now creates a copy of the team configuration before mutation to avoid corrupting the original team object, ensuring safer and more predictable behavior. [1] [2]
• In the request processing workflow, now always forces orchestration rebuild for each new task to handle non-reusable Magentic workflows.

Code Cleanup and Legacy Removal:

• Removed unused imports and legacy functions related to agent database retrieval and storage, reflecting the new agent management model. [1] [2] [3] [4]
• Improved streaming agent response callback to handle various update object shapes more robustly and removed reliance on specific internal types. [1] [2]

These changes collectively modernize the backend to leverage the latest Azure AI SDKs, improve maintainability, and ensure more robust agent orchestration and logging.

Does this introduce a breaking change?

• Yes
• No

How to Test

• Get the code

git clone [repo-address]
cd [repo-name]
git checkout [branch-name]
npm install

• Test the code

What to Check

Verify that the following are valid

• ...

Other Information