skills/klotzkette/claude-fuer-deutsches-recht/werkvertrag-dienstvertrag-software/Gen Agent Trust Hub
werkvertrag-dienstvertrag-software
Pass
Audited by Gen Agent Trust Hub on Sep 30, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes user-provided legal documents, case facts, and contract texts, which represents a potential surface for indirect prompt injection where malicious instructions could be embedded in data. However, this is inherent to the skill's primary purpose of legal document analysis.
- Ingestion points: Processes external data mentioned in Workflow (Sachverhalt zerlegen) and Arbeitsweg (Dokumente und Beweismittel).
- Boundary markers: Not explicitly defined in the skill instructions.
- Capability inventory: No dangerous capabilities (such as command execution or file writes) are requested or defined within the skill.
- Sanitization: The skill does not specify input sanitization but emphasizes manual verification of facts and legal sources.
- [SAFE]: The external resources mentioned, such as gesetze-im-internet.de, dejure.org, and openJur, are established and trusted public legal databases. The skill correctly instructs the agent to verify model-generated knowledge against these official sources.
Audit Metadata