Content Moderation with Logging – Additive Callbacks

Demonstrates before_model and after_model callbacks. The scenario: a content moderation agent where we log every request before it reaches the model and audit every response after generation.

Tip

What you’ll learn How to register lifecycle callbacks with accumulation semantics.

Source: 03_callbacks.py

adk-fluent

from adk_fluent import Agent

agent_fluent = (
    Agent("content_moderator")
    .model("gemini-2.5-flash")
    .instruct(
        "You are a content moderation agent. Evaluate user-submitted "
        "content and flag anything that violates community guidelines. "
        "Provide a severity rating: safe, warning, or violation."
    )
    .before_model(log_moderation_request)
    .after_model(check_response_safety)
    .build()
)

Native ADK

from google.adk.agents.llm_agent import LlmAgent


def log_moderation_request(callback_context, llm_request):
    """Log incoming content for audit trail before the model processes it."""
    print("[AUDIT] Moderation request received")


def check_response_safety(callback_context, llm_response):
    """Verify model output meets safety standards after generation."""
    print("[AUDIT] Response safety check passed")


agent_native = LlmAgent(
    name="content_moderator",
    model="gemini-2.5-flash",
    instruction=(
        "You are a content moderation agent. Evaluate user-submitted "
        "content and flag anything that violates community guidelines. "
        "Provide a severity rating: safe, warning, or violation."
    ),
    before_model_callback=log_moderation_request,
    after_model_callback=check_response_safety,
)

Equivalence

assert type(agent_native) == type(agent_fluent)
assert agent_native.before_model_callback == agent_fluent.before_model_callback
assert agent_native.after_model_callback == agent_fluent.after_model_callback