external-gitcode-ascend-model-training

Pass

Audited by Gen Agent Trust Hub on May 19, 2026

Risk Level: SAFE
Full Analysis
  • [SAFE]: The skill describes a standard workflow for AI training on Ascend NPUs.- [COMMAND_EXECUTION]: The skill utilizes standard shell commands (bash, ps, grep, tail, pkill) and environment variable exports necessary for hardware-accelerated model training. The usage of hostname -I is standard for identifying network interfaces in distributed training environments (HCCL).- [EXTERNAL_DOWNLOADS]: No external downloads or remote code executions were detected. The skill references local scripts within defined working directories. The source repository is a well-known platform for the hardware ecosystem mentioned.- [INDIRECT_PROMPT_INJECTION]: The skill uses placeholders such as <num_gpus> and <model_directory> to be populated during interaction. While this represents a theoretical interface for command injection if input is not sanitized, it is a standard pattern for developer-oriented training tools.
Audit Metadata
Risk Level
SAFE
Analyzed
May 19, 2026, 06:27 AM
Security Audit — agent-trust-hub — external-gitcode-ascend-model-training