Short answer: Yes. In Prefect 3.x, setting -

PREFECT_RESULTS_PERSIST_BY_DEFAULT=true

, and -

PREFECT_DEFAULT_RESULT_STORAGE_BLOCK=s3-bucket/my-s3-block

(or the canonical

PREFECT_RESULTS_DEFAULT_STORAGE_BLOCK=s3-bucket/my-s3-block

) will cause flow and task results to be persisted by default to that S3 block, unless you explicitly override persistence on a given flow/task. Details and gotchas: - Either env var name for the storage block works:

PREFECT_RESULTS_DEFAULT_STORAGE_BLOCK

(recommended) or

PREFECT_DEFAULT_RESULT_STORAGE_BLOCK

(also supported). - Precedence rules: - Task/flow-level

persist_result

or

result_storage

parameters override the env vars. - If a task has

@task(persist_result=False)

, it will not persist even with the env vars set. - The S3 block must exist in your workspace and be loadable from the run environment. - Make sure the runtime has: - The Prefect AWS collection installed (

prefect-aws

) so the S3 block can write to S3. - IAM permissions to write to the target bucket (e.g., via the CodeBuild role or AWS_* env vars). - Optional (not required): If you want non-pickle serialization, set

PREFECT_RESULTS_DEFAULT_SERIALIZER=json

. No other Prefect env vars are strictly required for default result persistence to S3 beyond the two you listed.

Great question. Think of two layers of defaults: - PREFECT_RESULTS_PERSIST_BY_DEFAULT (global default for flows and tasks) - PREFECT_TASKS_DEFAULT_PERSIST_RESULT (task-only default that overrides the global default) What PREFECT_TASKS_DEFAULT_PERSIST_RESULT does: - It sets a workspace-wide default for all tasks’ persistence behavior. - It only applies when a task doesn’t have an explicit

persist_result

set. - It takes precedence over both the flow’s

persist_result

and

PREFECT_RESULTS_PERSIST_BY_DEFAULT

. Precedence for a task’s persistence (highest to lowest): 1) Task-level

persist_result

on the decorator 2) PREFECT_TASKS_DEFAULT_PERSIST_RESULT 3) Flow-level

persist_result

4) PREFECT_RESULTS_PERSIST_BY_DEFAULT 5) Default: False (no persistence) Concrete examples: - If you set

PREFECT_TASKS_DEFAULT_PERSIST_RESULT=false

and your flow has

@flow(persist_result=True)

, tasks will NOT persist (the task default overrides the flow). - If you set

PREFECT_RESULTS_PERSIST_BY_DEFAULT=true

but do not set

PREFECT_TASKS_DEFAULT_PERSIST_RESULT

, tasks will persist unless the flow/task says otherwise. - If a task has

@task(persist_result=True)

, it will persist regardless of either env var. Env var name: - The correct env var is

PREFECT_TASKS_DEFAULT_PERSIST_RESULT

(no aliases).