health-score-design
Pass
Audited by Gen Agent Trust Hub on Nov 8, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONNO_CODE
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes external data files such as segmentation models and churn history (via
--segmentationand--churn-historyarguments). While it lacks boundary markers for these inputs, the risk is minimized by strict instructional guardrails (e.g., hard floors for behavioral weights and ceilings for sentiment weights) that govern how the agent interprets and applies the data. - [NO_CODE]: The skill consists entirely of instructional markdown and logical frameworks. It does not include any shell scripts, binary executables, or dynamic code execution blocks, significantly reducing its attack surface.
Audit Metadata