Informative Priors for MMM
Pass
Audited by Gen Agent Trust Hub on Oct 23, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [SAFE]: The skill provides instructional content and code for statistical modeling. Analysis of the instructions and Python snippets did not reveal any malicious patterns, such as command injection, unauthorized network access, obfuscation, or credential harvesting. The activities described align with standard data science workflows for model parameterization.
- [INDIRECT_PROMPT_INJECTION]: The skill includes code to read from a local data file ('cleaned_data.parquet') in SKILL.md, which represents a data ingestion surface. However, the subsequent processing is limited to statistical calculations and model configuration within the PyMC framework. There is no evidence of unsafe interpolation of external content into control structures or system prompts that would lead to unauthorized instruction execution.
Audit Metadata