microsoft/semantic-kernel · error · ValueError

The service must support structured output.

Error message

The service must support structured output.

What it means

Raised by get_response_from_structured_output_tool when the supplied chat completion service's prompt execution settings object has no response_format attribute, meaning the service cannot emit structured/schema-validated output. The helper relies on response_format to force the model to produce JSON conforming to the target pydantic schema, so a service lacking it is unsupported.

Source

Thrown at python/semantic_kernel/agents/orchestration/tools.py:42

    Args:
        target_structure (type): The target structure to transform the output into.
        service (ChatCompletionClientBase): The chat completion service to use for the transformation. This service
            must support structured output.
        prompt_execution_settings (PromptExecutionSettings, optional): The settings to use for the prompt execution.

    Returns:
        Callable[[DefaultTypeAlias], Awaitable[BaseModel]]: A function that takes the output of
            the chat completion service and transforms it into the target structure.
    """
    kernel = Kernel()
    kernel.add_service(service)

    settings = kernel.get_prompt_execution_settings_from_service_id(service.service_id)
    if prompt_execution_settings:
        settings.update_from_prompt_execution_settings(prompt_execution_settings)
    if not hasattr(settings, "response_format"):
        raise ValueError("The service must support structured output.")
    settings.response_format = target_structure

    chat_history = ChatHistory(
        system_message=(
            "Try your best to summarize the conversation into structured format:\n"
            f"{target_structure.model_json_schema()}."
        ),
    )

    async def output_transform(output: DefaultTypeAlias) -> BaseModel:
        """Transform the output of the chat completion service into the target structure."""
        if isinstance(output, ChatMessageContent):
            chat_history.add_message(output)
        elif isinstance(output, list) and all(isinstance(item, ChatMessageContent) for item in output):
            for item in output:
                chat_history.add_message(item)
        else:
            raise ValueError(f"Output must be {DefaultTypeAlias}.")

View on GitHub (pinned to c028a0c7dc)

Solutions

1. Use a service that supports structured output, e.g. AzureChatPromptExecutionSettings-based services (AzureChatCompletion) or OpenAIChatCompletion.
2. Upgrade the semantic_kernel connector to a version that exposes response_format on the service's settings.
3. If your service only supports tool/function calling, use that path instead of get_response_from_structured_output_tool.
4. Wrap a custom service in a subclass whose settings declare response_format.

Example fix

// before
from semantic_kernel.connectors.ai.open_ai import OpenAIChatCompletion
service = OpenAIChatCompletion(service_id="local", ai_model_id="x", ...)  # settings w/o response_format
tool = get_response_from_structured_output_tool(service, MyModel)
// after
from semantic_kernel.connectors.ai.open_ai import AzureChatCompletion
service = AzureChatCompletion(service_id="az", deployment_name="gpt-4o", endpoint=..., api_key=...)
tool = get_response_from_structured_output_tool(service, MyModel)

from semantic_kernel import Kernel

def service_supports_structured_output(service) -> bool:
    kernel = Kernel(); kernel.add_service(service)
    settings = kernel.get_prompt_execution_settings_from_service_id(service.service_id)
    return hasattr(settings, "response_format")

def has_response_format(service) -> bool:
    try:
        from semantic_kernel import Kernel
        k = Kernel(); k.add_service(service)
        return hasattr(k.get_prompt_execution_settings_from_service_id(service.service_id), "response_format")
    except Exception:
        return False

try:
    tool = get_response_from_structured_output_tool(service, MyModel)
except ValueError as e:
    if "structured output" in str(e):
        service = AzureChatCompletion(...)  # switch to a supporting service
        tool = get_response_from_structured_output_tool(service, MyModel)
    else:
        raise

When it happens

Trigger: Calling get_response_from_structured_output_tool(service, target_structure, ...) with a service whose settings class does not define response_format (e.g. a non-OpenAI service, a custom AzureOpenAI config missing the field, or an older/limited connector).

Common situations: Swapping an OpenAI service for a local/Ollama/Azure variant that lacks response_format support, using a chat completion service class that predates structured output support, or passing a service that only supports function-calling (tools) but not response_format.

Related errors

• Output must be {DefaultTypeAlias}.
• Unable to transform result into {typeof(TOutput).Name}
• A complete listen_for condition is required for orchestratio
• At least one then action is required for orchestration steps
• The service must support structured output.

AI-assisted analysis of microsoft/semantic-kernel@c028a0c7dc (2026-08-13). Data as JSON: /api/errors/2354ee6bd80444b3. Report an issue: GitHub.