nemo-mbridge-perf-moe-long-context

Pass

Audited by Gen Agent Trust Hub on Jul 22, 2026

Risk Level: SAFE
Full Analysis
  • [SAFE]: The skill provides technical guidance for training MoE models, specifically covering context parallelism (CP) sizing and selective recompute strategies. No executable scripts, shell commands, or automation tools were found within the skill files.- [SAFE]: No network operations, data exfiltration patterns, or hardcoded credentials were detected. All external references are limited to documentation paths and performance tuning guides in standard markdown format.- [SAFE]: The skill does not perform any package installations or remote code execution. The configuration parameters listed (such as TP, CP, EP, and PP values) are intended as manual inputs for large-scale training recipes.
Audit Metadata
Risk Level
SAFE
Analyzed
Jul 22, 2026, 07:46 AM
Security Audit — agent-trust-hub — nemo-mbridge-perf-moe-long-context