The Agent Skills Directory

[SAFE]: The skill operates entirely locally, parsing LaTeX and repository metadata to ensure reproducibility. It does not utilize network protocols or access sensitive system directories.\n- [SAFE]: Security guardrails are explicitly defined in the instructions, ensuring that the AI agent only provides guidance while the human user retains control over environment modification and heavy code execution.\n- [SAFE]: All provided scripts use Python standard libraries and perform focused parsing tasks without dangerous dynamic execution sinks like eval() or subprocess calls.\n- [SAFE]: External URLs referenced in documentation point to well-known academic services (e.g., ACM, Zenodo) for informational purposes and do not pose a security risk.

verify-results