nemo-mbridge-perf-moe-dispatcher-selection
Pass
Audited by Gen Agent Trust Hub on Jul 22, 2026
Risk Level: SAFENO_CODE
Full Analysis
- [SAFE]: The skill was analyzed for prompt injection, data exfiltration, obfuscation, and other malicious patterns. No security risks or suspicious behaviors were detected.
- [NO_CODE]: The skill consists entirely of markdown documentation, YAML metadata, and evaluation results. It does not include any executable scripts, binaries, or code snippets that the agent would execute on the host system.
- [COMMAND_EXECUTION]: The skill provides examples of command-line flags (e.g.,
--moe-deepep-num-sms,--moe-hybridep-num-sms) intended for use with external training recipes. These are provided as static documentation for user reference and do not involve direct shell execution or command injection vulnerabilities.
Audit Metadata