Short answer: yes—no monkeypatching needed. In Prefect 3.x, the clean options in pytest are: 1) Run the task inside a flow (context is created for you) - Easiest way to test logging from tasks; just capture with pytest’s caplog.

from prefect import flow, task
from prefect.logging import get_run_logger

def test_task_logs(caplog):
    @task
    def t():
        get_run_logger().info("hello from task")

    @flow
    def f():
        t()

    f()
    assert "hello from task" in caplog.text

2) Create a minimal run context for unit tests (no orchestration) - If you want to call the task’s function body directly (

task.fn()

), wrap it in a minimal FlowRunContext to avoid MissingContextError.

from prefect import task
from prefect.context import FlowRunContext
from prefect.logging import get_run_logger

def test_task_fn_with_minimal_context(caplog):
    @task
    def t():
        get_run_logger().info("hello from unit test")

    # Minimal context: enough for get_run_logger to work
    with FlowRunContext.model_construct(flow=None, flow_run=None):
        t.fn()  # call the raw function body

    assert "hello from unit test" in caplog.text

- You can also pass the context explicitly:

ctx = FlowRunContext.model_construct(flow=None, flow_run=None)
logger = get_run_logger(context=ctx)
<http://logger.info|logger.info>("hello via explicit context")

3) If you don’t care about logging in a test, disable it - This avoids raising when there’s no run context.

from prefect import task
from prefect.logging import disable_run_logger, get_run_logger

def test_task_no_logging_needed():
    @task
    def t():
        # Will return a disabled logger instead of raising
        get_run_logger().info("won't be emitted")

    with disable_run_logger():
        t.fn()

Notes -

get_run_logger()

raises MissingContextError if called outside a flow/task run context. Running the task within a flow or providing a minimal FlowRunContext is the neat fix. - If you specifically need task-run metadata (IDs, names), the simplest path is to run the task via a flow (option 1). Manually building a full TaskRunContext is possible but generally more work for unit tests. If you can share a tiny snippet of how you’re testing now, I can suggest the minimal change that removes your monkeypatch.