Ability to get Stack Outputs as a Data Source

[ikegentz]

It would be great to have a way to get the outputs from other stacks with a data source.

For example:

data "spacelift_stack_output" "other_stack_output" {
  stack = "some other stack"
  key   = "output key"
}

Right now I'm not aware of a way to do this short of running a null_resource and querying the graphQL API directly.

[marcinwyszynski]

Spacelift provides a dedicated way to consume outputs of stacks, in the form of references.
Are there use cases that would be better covered by the hypothetical spacelift_stack_output data source?

[mbainter]

For us, references would be the reason for wanting the resource.

Consider a scenario where you have a stack that is creating the environment for a service, and you want to make that information available via a context that can attach to any service in the space that will be deploying to that environment.

In order to do this, you have to create an individual reference for every output that you want to connect between those stacks. Later, if you go back and say, add a bunch of additional subnets, then you have to also create more reference connections between those stacks. You can, of course, use foreach and a list(string) in terraform to DRY this up, but you will still need to manually maintain that list of outputs.

If you could get a list of all of the available outputs, then you wouldn't have to do that. You could just iterate on that datasource and create references automatically.

[amir-khatibzadeh]

For teams managing multiple stacks, referencing outputs from one stack in another is a common need. A use case includes dynamically passing environment details between stacks without requiring manual reference creation for every new output.

Currently, users must explicitly define references for each output they wish to connect. If additional outputs (e.g., new subnets) are added later, they must also manually update reference connections. While foreach and list(string) can help, users must still maintain the output list manually.

If a data source provided a list of all available outputs, it would enable automatic iteration and reference creation, simplifying stack dependencies.

Implementation
I have developed two Terraform data resources that address this functionality, available on both the Terraform and OpenTofu registries:
📄 Documentation: Terraform Registry

[jstrese]

Managing Spacelift with Terraform is resource heavy. There are many resources needed to set up even one stack. Now multiply that by how many environments you have and it really adds up.

Instead of relying on explicit data references, I would like to see Spacelift support variables in their values for input variables.

Example:
StackA outputs vpc_id

StackB needs vpc_id from StackA

StackB creates an input variable named TF_VAR_vpc

The value is set to
output_ref:<space|optional>.StackA.vpc_id

Spacelift would then pull the value at apply time and/or manage the reference behind the scenes.

If it can't be resolved, throw an error at apply/plan phase.

Or implement something like autoattach contexts.