deterministic-metric-design
Pass
Audited by Gen Agent Trust Hub on Jul 6, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill's primary function is as a reference and methodological framework. It contains Markdown-based instructional content without executable components that would pose a security risk.
- [SAFE]: External references are limited to established academic sources (ACM, IEEE, Springer), scientific journals (Science), and official technical documentation (NIST, Python docs, Scikit-learn). These links are clearly visible and serve to provide the theoretical basis for the metric design rules.
- [SAFE]: Code snippets provided in the rule files are intended as illustrative examples of metric calculation logic (e.g., using AST parsing or statistical summaries) and do not contain instructions for the agent to perform dangerous operations or access sensitive data.
- [SAFE]: The skill does not request or require any third-party dependencies for runtime execution. Mentions of pnpm and local validation scripts are directed at developers maintaining the repository, not at the agent during typical skill invocation.
Audit Metadata