external-gitcode-ascend-triton-operator-dev

Pass

Audited by Gen Agent Trust Hub on May 19, 2026

Risk Level: SAFE
Full Analysis
  • [SAFE]: The skill serves as a high-level workflow manager and documentation guide. It provides a structured approach to development and does not contain any malicious instructions, obfuscation, or unauthorized access attempts.
  • [EXTERNAL_DOWNLOADS]: The metadata references a synchronization source on GitCode (gitcode.com/Ascend/agent-skills). GitCode is a recognized developer platform, and the Ascend organization is a well-known entity in the AI hardware sector. This reference is for metadata and provenance and does not involve automated malicious downloads.
  • [COMMAND_EXECUTION]: The skill describes a development workflow that involves executing standard profiling tools (msprof) and Python verification scripts. These operations are essential to the skill's stated purpose of operator development and evaluation.
  • [PROMPT_INJECTION]: The instructions focus entirely on the technical stages of operator development. There are no patterns suggesting attempts to bypass safety filters, override system prompts, or manipulate agent behavior.
  • [DATA_EXFILTRATION]: No instructions or scripts were found that access sensitive local files (e.g., credentials, SSH keys) or transmit data to unauthorized external endpoints.
Audit Metadata
Risk Level
SAFE
Analyzed
May 19, 2026, 06:27 AM
Security Audit — agent-trust-hub — external-gitcode-ascend-triton-operator-dev