rocm-kernels

MEDIUM W011: Third-party content exposure detected (indirect prompt injection risk).

• Third-party content exposure detected (high risk: 0.90). The skill explicitly loads public, third‑party model and kernel artifacts (e.g., Lightricks/LTX-Video via LTXPipeline.from_pretrained in SKILL.md and scripts/benchmark_e2e.py, and community kernels via get_kernel/has_kernel in references/huggingface-kernels-integration.md and scripts), and those downloaded user/Hub-provided modules are executed/used at runtime to patch and drive model behavior, so untrusted content can materially change agent actions.