perf-moe-dispatcher-selection

Pass

Audited by Gen Agent Trust Hub on May 25, 2026

Risk Level: SAFE
Full Analysis
  • [SAFE]: The skill is a documentation-only resource providing architectural guidance for MoE (Mixture-of-Experts) model training.
  • [COMMAND_EXECUTION]: The provided commands are restricted to example CLI flags (e.g., --moe-deepep-num-sms) and environment variables (e.g., CUDA_DEVICE_MAX_CONNECTIONS) used for GPU kernel optimization. These are standard parameters for high-performance computing and do not present a security risk.
  • [NO_CODE]: No scripts, binaries, or external dependencies are included. The analysis found no evidence of network activity, file system manipulation, or credential access.
Audit Metadata
Risk Level
SAFE
Analyzed
May 25, 2026, 02:31 AM
Security Audit — agent-trust-hub — perf-moe-dispatcher-selection