Is there a way to get some coverage for this by using it in our own tests? E.g. somehow cleaning up this pattern?

async def test_some_test(client: Client):
    new_config = client.config()
    new_config["plugins"] = [
        openai_agents.OpenAIAgentsPlugin(
            model_params=ModelActivityParameters(
                start_to_close_timeout=timedelta(seconds=30)
            ),
            model_provider=CustomModelProvider(),
        )
    ]
    client = Client(**new_config)

Is there a way to get some coverage for this by using it in our own tests? E.g. somehow cleaning up this pattern?

Potentially. I might have to understand more about pytest fixtures though, and it might be more difficult to see what is actually being validated. The same could theoretically be said for any other client modifications we do.