nemo-rl-auto-research

Pass

Audited by Gen Agent Trust Hub on Jun 12, 2026

Risk Level: SAFECOMMAND_EXECUTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill guides the agent to perform repository actions (git branching, git commit) and launch training experiments (uv run, launch-nemo-rl). These actions are aligned with the skill's primary purpose of automating research and include safety instructions to seek user confirmation before execution.- [EXTERNAL_DOWNLOADS]: The skill utilizes uv for Python package management and references official NVIDIA documentation and tools. These references are considered safe and typical for the vendor's ecosystem.- [SAFE]: No malicious patterns such as obfuscation, credential harvesting, or unauthorized network access were found. The skill explicitly directs the agent to seek user confirmation before starting long-running jobs or making significant repository changes. Minor best-practice violations like missing metadata fields (author, tags) do not impact the security of the skill.- [PROMPT_INJECTION]: The instructions provide a clear operational scope and include constraints (e.g., "Do NOT use for: bug fixes") that help prevent the agent from deviating into unintended or unsafe tasks.
Audit Metadata
Risk Level
SAFE
Analyzed
Jun 12, 2026, 07:42 PM
Security Audit — agent-trust-hub — nemo-rl-auto-research