# Prefect version, path: clients/GOOGLE/test_flow/prefect.yaml
prefect-version: 3.2.9  # Replace with the Prefect version you're using
name: test-flow

pull:
- prefect.deployments.steps.git_clone:
    id: clone-step   # needed to be referenced in subsequent steps
    repository: <https://github.com/user/codebase.git>
    credentials: '{{ prefect.blocks.github-credentials.github-credential-user }}'
- prefect.deployments.steps.pip_install_requirements:
    id: requirements-step   
    directory: clients/GOOGLE/test_flow
    requirements_file: requirements.txt

# Deployment configurations
deployments:
- name: test-flow-github-deploy
  entrypoint: flows/test_flow.py:test_flow
  work_pool:
    name: default-pool
    work_queue_name:
    job_variables: {}
  schedule:
  parameters: 
  version: '0.1'
  tags: []
  concurrency_limit:
  description: 
  schedules: []

my run fails because its expecting

flows/test_flow.py

to be at root of repo where as it is inside

clients/GOOGLE/test_flow

, the same location as

requirements.txt

and my

prefect.yaml

for this flow deploy. if I change entrypoint to ``clients/GOOGLE/test_flow/flows/test_flow.py`` then the deploy fails since it runs entrypoint relative to prefect.yaml path. here is the dir struct:

repo_root:
- clients/GOOGLE/test_flow/flows
- clients/GOOGLE/test_flow/tasks
- clients/GOOGLE/test_flow/prefect.yaml
- clients/GOOGLE/test_flow/requirements.txt
- clients/YAHOO/test_flow/flows
- clients/YAHOO/test_flow/tasks
- clients/YAHOO/test_flow/prefect.yaml
- clients/YAHOO/test_flow/requirements.txt

endgoal is to use one repo to manage all clients flows, how do I set directory inside deployments if my yaml files are inside each client folders?

hi @Victor - have you tried using the

--prefect-file

flag for

prefect deploy

? you could check out this example https://github.com/zzstoatzz/prefect-pack/tree/main/flows/be_modal where im following a couple hard rules • always run

prefect deploy

from repo root • always specify

entrypoint

relative to repo root because this example was motivated by a user who wanted to run

make deploy

from within their specific folder, the

make

command cd's out to root you might want to plus 1 this issue, as it would relax that second rule above and perhaps make setups like yours easier to reason about

Hey Nate, thank you for such a detailed answer.

-prefect-file

arg worked but was facing issues with tasks import failure in flows due to path issues. I was able to add

___init___.py

to all folders to make them python module and then refer to tasks folder using root reference. Thank you so much for the help.

my flow fails with errors tasks not found if I do this in my flow:

from tasks.download_reports import generate_and_download_pdfs

but if I update it to:

from clients.GOOGLE.test_flow.tasks.download_reports from generate_and_download_pdfs

it stops working locally and I am unable to test things. same goes with my utils folder.

i think the solution here is to build a proper python package, for example: https://github.com/zzstoatzz/prefect-pack/blob/main/pyproject.toml

» uv sync
Resolved 136 packages in 1ms
Audited 132 packages in 0.28ms
» uv run python -c "from prefect_pack.utils import parse_as;print(parse_as)"
<function parse_as at 0x104e172e0>

so i'm using

uv

(which I'd recommend) but you can use

pip install .

as i mention in the readme and this way you have a static and unambiguous import,

from your_package.submodule import your_thing

so whether you do the

pip install .

in your dockerfile or in a

pull

step or something, that's up to you, but I would build a python package so your imports are consistent