llm-inference-scaling
Pass
Audited by Gen Agent Trust Hub on Sep 15, 2026
Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADS
Full Analysis
- [COMMAND_EXECUTION]: The skill includes legitimate shell commands for cluster management and infrastructure deployment.
- Evidence: Provides
helmandkubectlcommands to install the NVIDIA GPU Operator and Cluster Autoscaler. - Evidence: Includes commands for verifying node status and querying Prometheus metrics.
- [EXTERNAL_DOWNLOADS]: The skill references standard infrastructure components from official and well-known repositories.
- Evidence: Downloads Helm charts from NVIDIA's official registry (
https://helm.ngc.nvidia.com/nvidia). - Evidence: References well-known container images including
vllm/vllm-openaiand the officialcluster-autoscaler. - [SAFE]: The skill follows security best practices for handling sensitive information in a Kubernetes environment.
- Evidence: Recommends using
secretKeyRefto inject theHUGGING_FACE_HUB_TOKENfrom a Kubernetes secret rather than hardcoding it in the deployment manifest.
Audit Metadata