aktenaufbereiter-strafrecht
Pass
Audited by Gen Agent Trust Hub on Sep 3, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted documents (PDF, Word, indictment files), which represents a surface for indirect instructions. This risk is managed through a documented evidence chain: 1. Ingestion points: Digitalized case files and indictments referenced in SKILL.md; 2. Boundary markers: The 'Anti-Hallucination-Regel' provides explicit instructions to ignore external biases and stick to file content; 3. Capability inventory: The skill is restricted to structured table generation with no scripts or subprocesses defined; 4. Sanitization: Source attribution (Band/Blatt) and contradiction checking across multiple indices (Methodik section).
- [EXTERNAL_DOWNLOADS]: The skill references official German legal domains (bundesverfassungsgericht.de) and well-known legal databases (dejure.org, openjur.de) for verifying legal information. These are authoritative sources and do not present a security risk.
Audit Metadata