def create_agent_from_yaml()

in generators/backend/templates/src/backend/utils/util.py [0:0]

• 21 lines of code
• 2 McCabe index (conditional complexity)

def create_agent_from_yaml(kernel, service_id, definition_file_path, reasoning_effort=None):
    """
    Creates a ChatCompletionAgent from a YAML definition file.
    
    Args:
        kernel: The Semantic Kernel instance
        service_id: The service ID to use for the agent
        definition_file_path: Path to the YAML file containing agent definition
        reasoning_effort: Optional reasoning effort parameter for OpenAI models
        
    Returns:
        ChatCompletionAgent: Configured agent instance
        
    The YAML definition should include name, description, instructions, 
    temperature, and included_plugins.
    """
        
    with open(definition_file_path, 'r', encoding='utf-8') as file:
        definition = yaml.safe_load(file)
        
    settings = AzureChatPromptExecutionSettings(
            temperature=definition.get('temperature', 0.5),
            function_choice_behavior=FunctionChoiceBehavior.Auto(
                filters={"included_plugins": definition.get('included_plugins', [])}
            ))

    # Resoning model specifics
    model_id = kernel.get_service(service_id=service_id).ai_model_id
    if model_id.lower().startswith("o"):
        settings.temperature = None
        settings.reasoning_effort = reasoning_effort
        
    agent = ChatCompletionAgent(
        service=kernel.get_service(service_id=service_id),
        kernel=kernel,
        arguments=KernelArguments(settings=settings),
        name=definition['name'],
        description=definition['description'],
        instructions=definition['instructions']
    )
    
    return agent