mamba-architecture
Pass
Audited by Gen Agent Trust Hub on Oct 23, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill provides instructions for installing well-known machine learning packages (
mamba-ssm,causal-conv1d,torch,transformers) from official package registries (PyPI). - [SAFE]: Remote model loading is performed using standard libraries (
transformers,mamba_ssm) from HuggingFace (state-spacesorganization), which is a trusted service for model weights. - [SAFE]: The benchmarking and training sections contain standard shell commands and Python code for model evaluation and distributed training without any indicators of command injection or malicious intent.
- [SAFE]: All external URLs link to official research papers (arXiv), the authorized GitHub repository (state-spaces/mamba), and official documentation, which are trusted sources for technical content.
Audit Metadata