nemo-mbridge-mlm-bridge-training

Pass

Audited by Gen Agent Trust Hub on Jul 16, 2026

Risk Level: SAFECOMMAND_EXECUTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill provides numerous shell commands for launching training jobs using uv run python -m torch.distributed.run. These are standard entry points for the Megatron-LM and Megatron Bridge frameworks.
  • [COMMAND_EXECUTION]: Provides instructions to switch the Megatron-Core submodule version using a local script ./scripts/switch_mcore.sh. This is a routine development task for managing internal dependencies.
  • [COMMAND_EXECUTION]: Includes a cleanup step rm -rf nemo_experiments to ensure fresh training runs do not resume from stale checkpoints. While this is a destructive command, it is constrained to a specific local project directory and is a documented requirement for correct training behavior.
Audit Metadata
Risk Level
SAFE
Analyzed
Jul 16, 2026, 11:54 PM
Security Audit — agent-trust-hub — nemo-mbridge-mlm-bridge-training