output_schema and tools cannot be used together in Agent configuration

[AlejandroBlanco2001]

Describe the bug

When creating agents, it is possible to specify both an output_schema and tools. However, the current documentation does not mention any restriction or limitation when using these two parameters together.

If both parameters are provided, a runtime error occurs. This is a known limitation: due to how output_schema is implemented, an agent cannot both enforce an output schema and call tools directly. The current expected workaround is to create a separate sub-agent responsible for calling the tool.

To Reproduce

Steps to reproduce the behavior:

1. Create an agent with the following code:

Agent(
    name="task_planner",
    model=create_litellm_model(model_name, api_base),
    instruction=(
        "You are a task planner. Return a JSON object with title, steps (array), "
        "and estimated_time (integer). Return ONLY valid JSON, nothing else."
    ),
    tools=[dummy_tool],
    output_schema=TaskPlan,
    output_key="task_plan",
)

Expected behavior

• A clear runtime error (or static type error) explaining that tools and output_schema cannot be used together.
• Updated documentation with a more explicit and visible note in the parameter descriptions explaining this limitation.

Screenshots

See the mentioned issue

Versions

• OS: macOS
• ADK version: 1.21
• Python version: 3.11

Additional context

See the following issue google/adk-python#3969 (comment)

[doitmaan]

is it done ?

[AlejandroBlanco2001]

@doitmaan I updated to the latest version, and the issue is not longer appearing for me

[doitmaan]

I'm glad to hear. Thanks for the update

[doitmaan]

I'm glad to hear. Thanks for the update