Yes — with the Docker worker you can enable automatic container cleanup using Docker’s auto-remove behavior. In Prefect 3.x, set the

auto_remove

job variable to

true

(this is equivalent to

docker run --rm

). By default it’s

false

, so containers stick around after the flow exits. Ways to enable it: - Per deployment (job variables) - Python deploy:

from prefect import flow

    @flow
    def my_flow():
        ...

    if __name__ == "__main__":
        my_flow.deploy(
            name="auto-cleanup",
            work_pool_name="my-docker-pool",
            job_variables={"auto_remove": True},
        )

- prefect.yaml:

deployments:
      - name: auto-cleanup
        work_pool:
          name: my-docker-pool
        job_variables:
          auto_remove: true

- Default for a Docker work pool (base job template) - In the UI: Work Pools -> your Docker pool -> Edit base job template -> set the

variables.auto_remove.default

to

true

and ensure

job_configuration.auto_remove

is wired to

{{ auto_remove }}

. - Or via CLI, inspect the current template:

prefect work-pool inspect my-docker-pool -o json

Then update the pool with a modified base job template. Notes and caveats: - This only removes the container; images, networks, and volumes are not pruned by Prefect. - With

auto_remove: true

, Docker deletes the container as soon as it exits. If you rely on inspecting containers after failure, leave it

false

. - If you still see leftover containers after enabling

auto_remove

, check whether a deployment-level

job_variables

is overriding the pool default. Want me to help you update your pool template or deployment config? Also, are you on Prefect 3.x and are the leftovers strictly containers (not volumes/images)?