amd-kernel-optimization

The AMD Kernel Optimization skill is coherently aligned with its stated purpose of guiding developers through a structured performance optimization workflow for ROCm-enabled PyTorch models on AMD GPUs. It relies on local, documented tooling and vendor libraries, with benchmarking and profiling as core activities and no evident need for unsafe downloads, credential handling, or external data exfiltration. The overall threat posture is low to moderate due to its reliance on standard, well-known components; no hidden data flows or credential harvesting patterns are detected. Given the lack of external dependencies, suspicious network activity, or credential exposure, the skill appears benign and appropriate for its claimed purpose.