PanickyDolphin50 the env var should be available where the agent is running and setting up the task - how are you running the agent, and in what mode?

Hi PanickyDolphin50 , what is this uv caching you're referring to?

You can add env variables to your tasks in the container section of the task

SuccessfulKoala55
i run agent in docker mode using that command clearml-agent daemon --queue baremetal --docker nvcr.io/nvidia/pytorch:23.05-py3 --detached

CostlyOstrich36
Its described here in pr None

I cant understand how to setup it properly on docker mode agent