mcore-run-on-slurm

Pass

Audited by Gen Agent Trust Hub on Jun 12, 2026

Risk Level: SAFE
Full Analysis
  • [SAFE]: The skill provides documentation and shell script templates for managing high-performance computing (HPC) workloads on SLURM clusters. The instructions follow established best practices for distributed training with PyTorch and Megatron-LM.
  • [COMMAND_EXECUTION]: The skill includes examples of SLURM job control commands such as sbatch, srun, squeue, and sacct. These are standard utilities for job scheduling and resource management in cluster environments and are required for the skill's stated purpose.
  • [EXTERNAL_DOWNLOADS]: The guide mentions using the uv package manager (uv sync) to prepare the Python environment and install dependencies. While this involves downloading packages, it is a standard development workflow for Megatron-LM, which is an official NVIDIA project.
  • [DATA_EXPOSURE]: The skill references environment variables such as MASTER_ADDR, MASTER_PORT, and SLURM_JOB_NODELIST. These are internal cluster networking parameters necessary for distributed coordination and do not involve the exposure of sensitive user credentials or private data.
Audit Metadata
Risk Level
SAFE
Analyzed
Jun 12, 2026, 07:41 PM
Security Audit — agent-trust-hub — mcore-run-on-slurm