Yes — there are a few practical conventions in Prefect 3.x, but not a heavy formal style guide beyond validation rules. Short version: - Use kebab-case for flow names and deployment names - Make names descriptive and stable - Treat

flow-name/deployment-name

as the important human-facing identifier - Avoid encoding too much changing metadata into the flow name itself What Prefect enforces Flow/deployment names cannot contain: -

/

-

%

-

&

-

>

-

<

So names like

daily-etl

are fine, but

daily/etl

is not. Recommended style Prefect docs and examples consistently use hyphen-separated lowercase names, e.g.: - Flows:

daily-etl

,

customer-sync

,

billing-reconciliation

- Deployments:

prod

,

staging

,

nightly

,

backfill

,

us-east

Also, if you do not explicitly name a flow, Prefect will turn a Python function like

my_flow_function

into

my-flow-function

, which is a good signal that kebab-case is the intended convention. Example:

python
from prefect import flow

@flow(name="daily-etl")
def run_daily_etl():
    pass

Then deployments might be:

python
run_daily_etl.from_source(...).deploy(
    name="prod",
    work_pool_name="my-pool",
)

That gives you a clear identifier:

text
daily-etl/prod

Good practice for naming A useful pattern is: - Flow name = what the workflow does - Deployment name = how/where/when it runs Examples: - Flow:

customer-sync

- Deployments:

prod

,

staging

,

hourly

,

manual-backfill

- Flow:

warehouse-load

- Deployments:

daily-prod

,

ad-hoc

,

dev

This is usually better than putting environment/schedule into the flow name itself unless they are truly different workflows. Recommended rules of thumb 1. Keep flow names stable - Prefer

customer-sync

- Avoid renaming frequently, since names are how people recognize runs in the UI/CLI 2. Put execution context in deployment names - Environment:

prod

,

staging

,

dev

- Trigger/schedule:

daily

,

hourly

,

manual

- Purpose:

backfill

,

smoke-test

3. Be concise but specific - Better:

stripe-invoice-sync

- Worse:

sync

- Worse:

data-pipeline-thing

4. Use one consistent format everywhere - Prefer all lowercase + hyphens - Avoid mixing

snake_case

,

camelCase

, and spaces 5. Don’t overstuff names -

customer-sync-prod-daily-us-east-batch-v2

is usually too much - If you need more metadata, use tags, parameters, descriptions, or work pools Suggested naming patterns Some patterns that work well: - By business function - Flow:

invoice-generation

- Deployments:

prod

,

staging

- By source -> destination - Flow:

salesforce-to-snowflake

- Deployments:

hourly

,

backfill

- By operation - Flow:

db-maintenance

- Deployments:

nightly

,

manual

When to create separate flows vs separate deployments Use separate deployments when the core logic is the same but execution context differs. Use separate flows when the workflows are meaningfully different. For example: - Same code, different schedule/env → one flow, multiple deployments - Different business process or task graph → separate flows Example I’d recommend

python
from prefect import flow

@flow(name="customer-sync")
def customer_sync():
    pass

Deployments: -

customer-sync/prod

-

customer-sync/staging

-

customer-sync/backfill

That tends to stay readable in both UI and CLI. Relevant docs/examples - Prefect flows concept docs: Flows - Prefect deployments docs: Deployments - Python deployment patterns with `from_source(...).deploy(...)`: Deploy via Python If you want, I can also suggest a team-wide naming convention template for flows + deployments for dev/staging/prod environments.

Yes — exactly.

/

is not valid inside the flow name or the deployment name itself, but Prefect uses

/

as the separator in the fully qualified deployment identifier:

text
<flow-name>/<deployment-name>

So this is valid:

text
customer-sync/prod

Where: - flow name =

customer-sync

- deployment name =

prod

But this would not be valid as a deployment name by itself:

text
prod/us-east

because

/

cannot appear inside the

name

field. So if you want that extra structure, prefer something like:

text
customer-sync/prod-us-east

or:

text
customer-sync/prod_us_east

though kebab-case like

prod-us-east

is the more common Prefect style. In practice: - Valid deployment reference:

my-flow/my-deployment

- Invalid flow name:

my/flow

- Invalid deployment name:

my/deployment

If helpful, I can also show where this matters in: -

prefect deploy

- Python

.deploy(name=...)

- running deployments from the CLI/API