skills/ascend-ai-coding/awesome-ascend-skills/external-gitcode-ascend-triton-operator-code-gen/Gen Agent Trust Hub
external-gitcode-ascend-triton-operator-code-gen
Pass
Audited by Gen Agent Trust Hub on May 19, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill provides technical instructions for implementing NPU kernels, including memory alignment, UB cache calculations, and core-specific tiling logic.
- [SAFE]: No network exfiltration, credential access, or malicious persistence mechanisms are present.
- [SAFE]: The code templates utilize standard libraries (torch, triton) required for Ascend NPU development.
- [PROMPT_INJECTION]: The skill is designed to ingest and parse operator requirement documents, which constitutes an indirect prompt injection surface. This is the primary function of the skill and is considered low risk as it does not expose exploitable capabilities within the instructions.
Audit Metadata