Suggestion / Question: Would it be possible to add...
# esc
b
Suggestion / Question: Would it be possible to add another option here where we can use the current_project and current_stack in the subject sent?
p
That is an interesting suggestion. Is it something you want to customize (https://www.pulumi.com/docs/pulumi-cloud/oidc/provider/#pulumi-esc:~:text=Additional%20options%20for%20customization%20include%3A) or just make it part of the default claims? @able-market-62580 - Thoughts?
a
It is interesting and actually discussed (particularly as part of anonymous environments and the OIDC subject). Problem is that ESC does not have a notion of the context where it is being executed (even in the case of the pulumi cli) so it is not straightforward to make in a reliable way to be consumed as part the provider authorization policies
b
I am wanting to lock down the access to environments by stacks. Right now we have rbac that covers team -> stack and team -> env. I want to further restrict to what stack can make use of it
I took a cruise through the code earlier and I am guessing that the deployments get a little special treatment behind the scenes since it all controlled on your end
a
sort of, only in that case we cold rely on the deployment token to carry that information in a trustworthy way, on the other hand, it worries me how inconsistent it will feel for customers not aware of these nuances.
b
yeah i actually misread i think because of the formating on that page,
(jumbled the deployments and ESC OIDC together
a
ah right, still you are talking about ESC right?
b
correct
i want to use ESC for OIDC
currently have team -> env and team -> stack
and looking to bridge the gap from stack -> env
a
yes makes sense
b
in all honesty i am being a bit paranoid anyway since if i give access to an env to a team they could do unintended thing in the stack the env was intended for anyway
a
true but it is a good usecase to track nonetheless (@plain-diamond-92898). I cant tell if we would be able to prioritize it in the short term but we discussed it a couple of times and disregarded due its complexity waiting for evidence that is needed
p
Totally. I think we should consider this as a case when we are thinking about upgrading our RBAC model.
Thanks for the feedback @brash-gigabyte-81569!
b
👍