datasets
Training and evaluation data, with the modality, task and licence stated up front. Listed live from the Hugging Face Hub.
tfplan-risk-labels-multicloud
Terraform Plan Risk Labels (Multi‑Cloud)
A dataset of Terraform plan JSON resource changes labeled with security/compliance risk categories (multi‑label) across AWS / GCP / Azure.
What it’s for
CI/CD pre-merge “risk scanning” of infrastructure changes
Fine-tuning / training a small classifier to predict risk labels from plan snippets
Bootstrapping deterministic OPA/Conftest policy packs (labels can be derived from policy denies)
Files
train.jsonl… See the full description on the dataset page: https://huggingface.co/datasets/bharathjanumpally/tfplan-risk-labels-multicloud.terraform-multicloud
terraform-multicloud
78,912 natural-language → Terraform (HCL) pairs, drawn from 25,264 public
GitHub repositories across 15 provider families. Every target is human-written
HCL that passes terraform validate. Every row carries its source repository,
licence and file path.
Built by the SASVA AI Model Cognition Labs (MCL) Team and used to train
SASVAAI/qwen38-27b-terraform.
What one row looks like
One complete row from data/train.jsonl, reproduced in full with… See the full description on the dataset page: https://huggingface.co/datasets/SASVAAI/terraform-multicloud.