datenschutzrecht-fehlerkatalog
Pass
Audited by Gen Agent Trust Hub on Aug 23, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill consists entirely of natural language instructions for auditing legal documents. It does not contain any scripts, command-line executions, or requests for network access. The mention of an external website (gesetze-im-internet.de) is a reference for human/model verification and not a programmed network call.
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to process user-provided legal documents ("Arbeitsergebnisse") to identify errors. While this represents an ingestion point for untrusted content, the skill lacks any capabilities that could be exploited by such an injection.
- Ingestion points: User-provided legal work results and drafts processed by the model.
- Boundary markers: None specified in the skill instructions.
- Capability inventory: No tools are allowed in the frontmatter, and no scripts or subprocess calls are present in the text.
- Sanitization: None defined for input, although the skill explicitly instructs the model to anonymize client data for privacy protection.
Audit Metadata