nemo-gym-debugging
Pass
Audited by Gen Agent Trust Hub on Nov 5, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: All components of the skill are dedicated to legitimate debugging and diagnostic workflows for the Nemo Gym framework. The scripts and command snippets are transparent and well-documented.
- [COMMAND_EXECUTION]: The skill provides diagnostic shell commands (e.g., using
rg,wc,stat) for inspecting log files and data outputs. These commands are intended for manual or agent-assisted troubleshooting of Slurm, Ray, and vLLM environments. - [DYNAMIC_EXECUTION]: Diagnostic references include Python snippets executed via heredocs. These scripts perform local data analysis, such as summarizing JSONL keys and checking task indices, and do not involve remote code execution or sensitive file access.
- [INDIRECT_PROMPT_INJECTION]: The skill processes external data (rollout logs and JSONL files). The analysis logic is limited to printing statistics and validation errors, which mitigates the risk of instructions within the processed data influencing the agent's behavior.
Audit Metadata