Hi, prefect team!! With prefect-server on k8s I want to create a Git-based flow. How do I install the pip packages required on the python flow in this case?

hi @tig - you have a couple options: • build an image for your flow with its dependencies and then provide this as the

image

on your k8s work pool, or as a

job_variables

override on

image

for that deployment like this • provide

EXTRA_PIP_PACKAGES

as an

env

field on your work pool like

EXTRA_PIP_PACKAGES="pandas numpy"

using either of the above methods • use a

run_shell_script

action in your pull step to run

pip install pandas numpy ...

in your

pull

step of your deployment

This way I can change the base pod settings used on the prefect-server side? I thought I had to use the docker based method and create an image for each flow and install the package there! Your input is very helpful!