kernel-tileir-optimization

Pass

Audited by Gen Agent Trust Hub on Aug 1, 2026

Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTION
Full Analysis
  • [SAFE]: The skill is designed for deep learning kernel optimization on Blackwell GPUs. The logic involves static analysis of Python code and templated string replacements to inject performance configurations.
  • [EXTERNAL_DOWNLOADS]: The skill references the official Triton-to-tile-IR repository on GitHub (github.com/triton-lang/Triton-to-tile-IR) for installing the nvtriton backend. These references target well-known open-source organizations and are documented as part of the standard setup process.
  • [COMMAND_EXECUTION]: The skill utilizes local Python scripts (scripts/classify_kernel.py, scripts/tileir_check.py) to automate development tasks. These scripts use standard Python libraries (argparse, re, json, importlib) to perform their functions without invoking dangerous shell commands or executing untrusted input.
  • [DATA_EXFILTRATION]: No evidence of sensitive file access or unauthorized network communication was found. Environment checks are limited to verifying GPU capabilities and package versions locally.
  • [PROMPT_INJECTION]: The instructions contain clear role definitions and operational guidelines for the agent. There are no attempts to bypass safety filters or override system instructions.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 1, 2026, 01:28 AM
Security Audit — agent-trust-hub — kernel-tileir-optimization