Multi-format IaC
Terraform, Pulumi, Crossplane, and Kubernetes manifests — generated and executed, not just sketched.
NebulaStack emits real, executable code — not a sketch you finish by hand. Terraform, Pulumi, and Crossplane are production-grade across all 12 providers. Kubernetes workloads get their own manifests, Helm charts, and Kustomize overlays.
- Terraform, Pulumi, and Crossplane — production-grade across all 12 providers
- Kubernetes manifests, Helm charts, and Kustomize overlays for container workloads
- A deterministic emitter writes the code — the language model doesn't, by default
How it works
A GenerateRequest is first mapped into a single canonical Intermediate Representation graph — content-addressed and schema-validated — before any output is written. That one graph then feeds separate emitters for Terraform HCL, Pulumi (TypeScript or Python), Crossplane compositions, and Azure Bicep, so every format describes the same infrastructure.
Per-resource fidelity warnings show where a format cannot fully represent something Terraform can. Pulumi and Crossplane track this per resource. They do not drop an attribute without telling you.
Terraform HCL
resource "aws_instance" "web" {
ami = "ami-0c94855ba95c71c99"
instance_type = "t3.medium"
}Pulumi (TypeScript) — same resource, same graph
const web = new aws.ec2.Instance("web", {
ami: "ami-0c94855ba95c71c99",
instanceType: "t3.medium",
});