Yes — programmatically, you can, but not via the older tag-limit helper API. For tag-based concurrency limits with slot decay, use the global concurrency limit API with a name like

tag:<your-tag>

. Example:

python
from prefect import get_client

async def main():
    async with get_client() as client:
        await client.upsert_global_concurrency_limit_by_name(
            name="tag:my-tag",
            limit=10,
            slot_decay_per_second=1.5,
        )

A few important notes: -

slot_decay_per_second

is supported on global concurrency limits - tag-based limits are represented as global limits named

tag:<tag-name>

- the legacy method

create_concurrency_limit(tag, concurrency_limit)

does not expose

slot_decay_per_second

If you want to update an existing one:

python
from prefect import get_client
from prefect.client.schemas.actions import GlobalConcurrencyLimitUpdate

async def main():
    async with get_client() as client:
        await client.update_global_concurrency_limit(
            name="tag:my-tag",
            concurrency_limit=GlobalConcurrencyLimitUpdate(
                slot_decay_per_second=2.0
            ),
        )

Conceptually: -

slot_decay_per_second = 0

-> standard concurrency limit -

slot_decay_per_second > 0

-> behaves like a rate limit, with slots decaying over time Relevant docs: - Global concurrency limits - Tag-based concurrency limits If you want, I can also show: - the sync client version - how to create this via the CLI - how this interacts with

@task(tags=[...])

usage