deepspeed

Pass

Audited by Gen Agent Trust Hub on Oct 29, 2026

Risk Level: SAFE
Full Analysis
  • [SAFE]: The skill serves as a technical documentation repository for the DeepSpeed optimization library. All content is educational and instructional in nature.
  • [COMMAND_EXECUTION]: The skill mentions administrative commands (e.g., apt install libaio-dev) and performance tuning options requiring elevated privileges (e.g., ds_nvme_tune --flush_page_cache). These are documented as necessary system requirements for the specific high-performance computing tasks DeepSpeed performs and are presented as technical guidance, not hidden or automated execution.
  • [EXTERNAL_DOWNLOADS]: The skill references official package registries and repositories (e.g., pip install deepspeed, git clone) for installing legitimate libraries. These are standard procedures for setting up the documented development environment.
  • [METADATA_POISONING]: The metadata correctly reflects the skill's purpose. The provided references and links point to official deepspeed.ai domains and associated GitHub organizations like microsoft, which are well-known and reputable services.
Audit Metadata
Risk Level
SAFE
Analyzed
Oct 29, 2026, 03:22 PM