Short answer: there isn’t a global “max runtime” switch in Prefect 3.x. You set timeouts per flow/task. If you need an org-wide default, the common pattern is to wrap Prefect’s

flow

decorator with your own. Options: - Flow-level timeout (recommended)

from prefect import flow

@flow(timeout_seconds=3600)  # 1 hour
def my_flow():
    ...

- Task-level timeout (for specific long tasks)

from prefect import task

@task(timeout_seconds=300)  # 5 minutes
def work():
    ...

- Granular timeout inside code

from prefect.utilities.timeout import timeout

with timeout(30):  # seconds
    do_something_blocking()

- Simulate a “global” default with a wrapper decorator

import os
from prefect import flow as prefect_flow

DEFAULT_FLOW_TIMEOUT = int(os.getenv("PREFECT_FLOW_TIMEOUT", "3600"))

def org_flow(*args, **kwargs):
    kwargs.setdefault("timeout_seconds", DEFAULT_FLOW_TIMEOUT)
    return prefect_flow(*args, **kwargs)

@org_flow
def my_flow():
    ...

What happens on timeout - The flow moves to a Failed state with a timeout error and Prefect attempts to interrupt execution. If you need cleanup, add hooks:

@flow(timeout_seconds=3600, on_cancellation=[...], on_failure=[...])
def my_flow():
    ...

Infrastructure-level notes - There’s no built-in “kill after T” at the worker level. If you run on Kubernetes, you can add

activeDeadlineSeconds

to your job/pod via your Kubernetes worker job template to hard-stop the container if you want an infra-enforced cap. See Workers and customize the Kubernetes job template accordingly. - For Docker/Process workers, rely on

timeout_seconds

or run under an external supervisor if you truly need a hard kill. Docs and references - Flow/task timeouts and timeout helpers: Timeout utilities - Hooks for cleanup: State change hooks - Cancellation concepts: Cancel workflows - Workers overview: Workers A couple of clarifying questions so I can tailor this: - Are you on Prefect 3.x, and which worker/infrastructure are you using (Process, Docker, Kubernetes)? - Do you want a default applied across all flows in a repo (code-level) or across your whole workspace (policy-level)?