Possible to suppress output events from LlmAgent as workflow nodes in ADK 2.0?

[markns]

I have some LlmAgents as workflow nodes roughly as follows:

classify_section = Agent(
    name="classify_section",
    model=_MODEL,
    instruction=CLASSIFY_SECTION_INSTRUCTION,
    output_schema=SectionPick,
    output_key="section_pick",
)

classify_heading = Agent(
    name="classify_heading",
    model=_MODEL,
    instruction=CLASSIFY_HEADING_INSTRUCTION,
    output_schema=HeadingPick,
    output_key="heading_pick",
)

def format_output(node_input):
    user_message = convert_to_user_message(node_input)
    return Event(output=user_message)

classifier_workflow = Workflow(
    name="hs_classifier",
    output_schema=HsClassificationResult,
    edges=[("START", classify_section, classify_heading, format_output)]

Ideally the user would only see an output event from the format_output node, but instead they see a serialised response from the classify_section and classify_heading agents, in addition to the format_output event.

Is it possible to suppress this output somehow?