skills/klotzkette/claude-fuer-deutsches-recht/alumni-stiftung-spenden-compliance/Gen Agent Trust Hub
alumni-stiftung-spenden-compliance
Pass
Audited by Gen Agent Trust Hub on Sep 3, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONNO_CODE
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill instructions direct the agent to process external data such as user-provided legal documents, case files, and correspondence. This creates a surface for indirect prompt injection where malicious instructions could be embedded in the analyzed data. However, the skill does not utilize any high-risk tools or capabilities that would allow for significant exploitation of this vector.
- Ingestion points: User-provided legal documents, protocols, and case facts (SKILL.md).
- Boundary markers: Absent; there are no instructions to differentiate between system guidelines and user-provided document content.
- Capability inventory: The skill performs text-based analysis and referencing; no file-writing, shell execution, or arbitrary network tools are invoked.
- Sanitization: Not present.
- [NO_CODE]: The skill contains natural language instructions and legal references only, with no executable scripts, system commands, or external code dependencies.
Audit Metadata