Skills
skills/pepperu96/hyper-mla/learn-cute-dsl/Gen Agent Trust Hub

learn-cute-dsl

Warn

Audited by Gen Agent Trust Hub on Apr 23, 2026

Risk Level: MEDIUMCOMMAND_EXECUTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill instructs the agent to execute shell commands for hardware profiling and performance testing, including nvidia-smi and the execution of benchmark kernels via python -m. It specifically directs the agent to run scripts/lock-gpu-clock.sh, a command that typically requires elevated system privileges to modify hardware performance states.
  • [COMMAND_EXECUTION]: The workflow involves generating executable Python source files (KernelPlan subclasses, __main__.py, and __init__.py) based on provided templates. These files are subsequently executed to perform correctness checks and benchmarking, representing a pattern of dynamic script generation and execution.
Audit Metadata
Risk Level
MEDIUM
Analyzed
Apr 23, 2026, 07:47 AM