behavioral-analysis
Pass
Audited by Gen Agent Trust Hub on Oct 16, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
- [SAFE]: The skill contains instructional content and code templates for analyzing political behavioral data within a defined parliamentary context. Behavior described (calculating voting deviation, group density, and leadership styles) aligns with the stated purpose of political intelligence.
- [INDIRECT_PROMPT_INJECTION]: The skill defines a vulnerability surface through the ingestion of untrusted data, specifically parliamentary documents and speeches (referenced in view_riksdagen_politician_document). The capability inventory is restricted to database read operations and statistical calculations. Analysis of the code shows that sanitization is handled via SQL parameter binding, though explicit prompt boundary markers for text processing are not defined in the snippets.
- [EXTERNAL_DOWNLOADS]: The skill references external documentation and policies hosted on GitHub and GitHub Pages. These references point to the vendor's own security policies and well-known developer documentation, representing standard informational linking.
Audit Metadata