urheberrechtliche-compliance
Pass
Audited by Gen Agent Trust Hub on Sep 30, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted user-provided content which presents an attack surface for indirect prompt injection.
- Ingestion points: The skill is designed to analyze user-supplied content such as "Originaldateien", "Promptlog", "KI-Output", "Verträge", and "Schriftsätze" as defined in SKILL.md.
- Boundary markers: The skill does not define explicit delimiters or "ignore embedded instructions" warnings to separate user-provided data from agent instructions.
- Capability inventory: The skill utilizes network capabilities to perform "live checks" of legal norms and case law on external portals.
- Sanitization: No evidence of data sanitization, validation, or escaping of the ingested content was found in the instructions.
- [EXTERNAL_DOWNLOADS]: The skill references and directs the agent to access official government and well-known legal databases such as gesetze-im-internet.de, dejure.org, openJur, and eur-lex.europa.eu for the purpose of verifying legal norms and case law.
Audit Metadata