deepspeed

Pass

Audited by Gen Agent Trust Hub on Jun 29, 2026

Risk Level: SAFE
Full Analysis
  • [SAFE]: The skill serves as a technical reference and educational guide for DeepSpeed developers. It contains no executable scripts or autonomous logic that could pose a security risk.
  • [COMMAND_EXECUTION]: The documentation includes various command-line examples (e.g., apt install libaio-dev, ds_nvme_tune, deepspeed --hostfile) and tool help outputs. These are provided for instructional purposes only and are consistent with official DeepSpeed usage patterns.
  • [EXTERNAL_DOWNLOADS]: Dependencies such as deepspeed, torch, and transformers, as well as external links to deepspeed.ai and associated GitHub repositories, point to official and well-known technology projects from trusted organizations (Microsoft/DeepSpeed).
Audit Metadata
Risk Level
SAFE
Analyzed
Jun 29, 2026, 12:55 AM
Security Audit — agent-trust-hub — deepspeed