skills/klotzkette/claude-fuer-deutsches-recht/angemessene-geheimhaltungsmassnahmen/Gen Agent Trust Hub
angemessene-geheimhaltungsmassnahmen
Pass
Audited by Gen Agent Trust Hub on Sep 3, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to process user-provided legal documents, creating an inherent but expected surface for indirect prompt injection.\n- Ingestion points: Ingests external documents like 'Verwaltungsakte' and 'Vertragsurkunden' for analysis as defined in the 'Arbeitsweg' section.\n- Boundary markers: Does not specify unique delimiters for user content, relying on the agent's contextual understanding to distinguish between legal facts and instructions.\n- Capability inventory: Limited to legal analysis, drafting document variants, and recommending related workflow skills within the agent's environment.\n- Sanitization: Relies on the agent to maintain 'Quellenhygiene' (source hygiene) by verifying citations against authoritative sources, though no programmatic sanitization of user input is defined.
Audit Metadata