abschlussprueferwahl
Pass
Audited by Gen Agent Trust Hub on Sep 3, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes external legal documents such as contracts, administrative acts, and expert opinions which constitutes an indirect prompt injection surface. However, the skill has no capabilities to write files, execute code, or perform network exfiltration, which significantly mitigates the risk. Ingestion points include 'Verwaltungsakte, Vertragsurkunden, Schriftsätze, Bescheide, Protokolle, Sachverständigengutachten' mentioned in SKILL.md. No specific boundary markers or sanitization are defined, but no dangerous capabilities are present.
- [SAFE]: The skill uses well-known legal databases like gesetze-im-internet.de and eur-lex.europa.eu for live norm verification, which is standard practice and considered safe behavior for this domain.
Audit Metadata