> ## Documentation Index
> Fetch the complete documentation index at: https://docs.fact0.io/llms.txt
> Use this file to discover all available pages before exploring further.

# Google ADK

> Instrument Google Agent Development Kit runs with telemetry callbacks.

# Google ADK

<Note>
  A native ADK extension is on the roadmap. The pattern below uses ADK's public callback hooks.
</Note>

## Install

```bash theme={null}
pip install fact0-sdk google-adk
```

## Pattern

Register `before_model_callback` / `after_model_callback` to emit spans.

```python theme={null}
import fact0
from google.adk.agents import LlmAgent
from google.adk.runners import Runner

client = fact0.Client(api_key="f0_live_...")

def before_model(ctx, request):
    ex = client.telemetry.current_execution()
    ctx.state["_lf_span"] = ex.start_span(
        "model.invoke", span_type="MODEL_INVOCATION",
    )

def after_model(ctx, response):
    span = ctx.state.pop("_lf_span", None)
    if span is not None:
        span.complete(output={"text": getattr(response, "text", "")[:500]})

agent = LlmAgent(
    name="researcher",
    model="gemini-2.0-flash",
    before_model_callback=before_model,
    after_model_callback=after_model,
)

with client.telemetry.execution(agent_id="adk.researcher") as ex:
    runner = Runner(agent=agent, app_name="research", session_service=...)
    for event in runner.run(user_id="u1", session_id="s1", new_message=...):
        ...
```

## Tool callbacks

ADK also exposes `before_tool_callback` and `after_tool_callback` - wrap them the same way and tag the span as `TOOL_CALL`.

## Related

* [Telemetry client](/sdk/python/telemetry)
* [OpenAI Agents SDK](/integrations/openai-agents-sdk)