So any code that needs the actual value of a resource should go inside the apply. Normally you only need to transform it to something else

repo_id = gitlab_repo.id.apply(lambda repo_id: f"dupa+{repo_id}")

repo_id

here is an

Output

object representing a value that, during the deployment, will be available and equal to the result of the lambda function The

pulumi.export("DUPA", repo_id)

don't export anything. It adds to the plan that after the deployment the values that was made available on that Output should be exported.