nemo-rl-brev-etiquette

Pass

Audited by Gen Agent Trust Hub on Jun 12, 2026

Risk Level: SAFE
Full Analysis
  • [DATA_EXFILTRATION]: The skill instructs the agent to load sensitive credentials (such as WANDB_API_KEY and HF_TOKEN) from a local environment file at /home/ubuntu/RL/.env. This is a standard and recommended practice for secret management.
  • [DATA_EXFILTRATION]: The instructions include a specific security constraint: 'Never print, cat, log, commit, or summarize secret values', which explicitly directs the agent to avoid common exfiltration or exposure patterns.
  • [COMMAND_EXECUTION]: The skill provides bash code blocks for checking disk space (df -h), creating directory structures (mkdir -p), and exporting environment variables. These operations are within the scope of the skill's intended purpose for environment setup.
  • [EXTERNAL_DOWNLOADS]: The skill-card references official NVIDIA documentation and GitHub repositories. These are verified vendor resources and do not involve untrusted third-party downloads.
  • [PROMPT_INJECTION]: The skill contains instructional content but no patterns associated with bypassing safety filters, ignoring instructions, or extracting system prompts.
Audit Metadata
Risk Level
SAFE
Analyzed
Jun 12, 2026, 07:41 PM
Security Audit — agent-trust-hub — nemo-rl-brev-etiquette