nemo-mbridge-perf-moe-comm-overlap

Pass

Audited by Gen Agent Trust Hub on Aug 11, 2026

Risk Level: SAFE
Full Analysis
  • [SAFE]: The skill consists of technical documentation and configuration recipes for training performance optimization. No malicious patterns such as prompt injection, data exfiltration, or obfuscation were detected.
  • [NO_CODE]: The skill files are composed of markdown and metadata (YAML/JSON). No executable scripts or binaries are bundled with the skill.
  • [COMMAND_EXECUTION]: The documentation provides standard bash command examples for running performance verification scripts (e.g., uv run python scripts/performance/run_script.py). These commands target local files within a project repository and do not involve remote execution or privilege escalation.
  • [CREDENTIALS_SAFE]: Analysis of all instructions and configuration flags confirms no sensitive data exposure, hardcoded secrets, or attempts to access private credential stores (like .ssh or .aws directories).
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 11, 2026, 04:02 PM
Security Audit — agent-trust-hub — nemo-mbridge-perf-moe-comm-overlap