[BUG] Tool is not called with structured output

[SZubarev]

Checks

• I have updated to the lastest minor and patch version of Strands
• I have checked the documentation and this is not expected behavior
• I have searched ./issues and there are no duplicates of my issue

Strands Version

1.8.0

Python Version

3.12.11

Operating System

macOS 15.6.1

Installation Method

pip

Steps to Reproduce

from strands import Agent, tool
from pydantic import BaseModel, Field

# Define a Pydantic model for the desired output structure
class UserInfo(BaseModel):
    id: str = Field(description="A unique identifier for the user")
    name: str = Field(description="The full name of the user")
    age: int = Field(description="The age of the user in years", ge=0, le=150)
    city: str = Field(description="The city where the user currently lives")


@tool
def generate_user_id() -> str:
    """Generate a unique user identifier

    Creates a universally unique identifier

    Returns:
        str: A unique user identifier
    """
    print("Generating user ID")
    return "User_12345"


def main():
    agent = Agent(
        model="us.anthropic.claude-sonnet-4-20250514-v1:0", tools=[generate_user_id]
    )

    print(agent.model.config)

    # Use structured_output to guide the agent to return a UserInfo object
    user_data = agent.structured_output(
        prompt="Generate information about random users. Use generate_user_id() tool to generate user identifier",
        output_model=UserInfo,
    )

    # user_data will be an instance of UserInfo, which can be easily converted to JSON
    print(user_data.model_dump_json())


if __name__ == "__main__":
    main()

Expected Behavior

We should see tool called and "User_12345" is used for UserID.

Actual Behavior

Random ID or "" is generated

Additional Context

No response

Possible Solution

No response

Related Issues

No response

[jwalaQ]

Structured output is not meant for tool calls. In the current implementation, structured_output has not been built to process prompts like a regular conversation. It is only meant to process an existing conversation and reframe it into the output model provided.

Structured output creates a tool called (in your case) "UserInfo" which is responsible only to process the current conversation and re-organize it into the fields. No other tool call is expected as part of the structured_output call.

The intended usage is

user_data = agent(
    prompt="Generate information about random users. Use generate_user_id() tool to generate user identifier"
)

structured_user_data = agent.structured_output(UserInfo, <optional prompt>)

Source

[SZubarev]

Thanks, it would be great to reflect it in documentation then.

[risan-raja-xebia]

Please include this in the documentation as well. As I am getting repeatedly ValueError: Stop Reason end_turn triggered.