llama-cpp
Pass
Audited by Gen Agent Trust Hub on Sep 8, 2026
Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSPRIVILEGE_ESCALATION
Full Analysis
- [COMMAND_EXECUTION]: The skill includes Python scripts that use
subprocess.runto callnvidia-smifor hardware profiling, specifically to calculate available VRAM for GPU layer offloading. - [EXTERNAL_DOWNLOADS]: The instructions guide the user to download source code from the official llama.cpp GitHub repository (
github.com/ggml-org/llama.cpp) and models from the Hugging Face Hub (huggingface.co). These are well-known and trusted services for AI development. - [PRIVILEGE_ESCALATION]: The troubleshooting guide mentions using
sudo apt installto install the NVIDIA CUDA toolkit, which is a standard administrative task for setting up GPU-accelerated computing environments.
Audit Metadata