llama-cpp
Pass
Audited by Gen Agent Trust Hub on Oct 8, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill serves as a comprehensive instructional resource for utilizing llama.cpp, covering installation, model conversion, and performance optimization.
- [EXTERNAL_DOWNLOADS]: The skill references downloads from well-known and reputable sources, such as the official llama.cpp GitHub repository and Hugging Face, for acquiring the tool's source code and pre-trained GGUF models.
- [COMMAND_EXECUTION]: The provided commands are standard for building software from source (e.g.,
make,brew install) and running the resulting binaries (./llama-cli,./llama-server) for local inference tasks. No high-risk or suspicious command patterns were detected. - [NO_CODE]: The skill does not include any executable scripts, drivers, or binary files of its own, relying entirely on the user's manual execution of documented commands and standard third-party tools.
Audit Metadata