Skip to content
#

kics

Here are 16 public repositories matching this topic...

Human-anchored security benchmark for LLM-generated Infrastructure-as-Code. 100 scenarios × 12 model configs = 1,196 artifacts scanned by Checkov/Trivy/KICS, compared against 634 human-written templates. Every model: 3.2–3.9× human vulnerability density. arXiv:2608.28021

  • Updated Aug 31, 2026
  • Python

Add this topic to your repo

To associate your repository with the kics topic, visit your repo's landing page and select "manage topics."

Learn more