aktuelle-lage-2026
Pass
Audited by Gen Agent Trust Hub on Sep 3, 2026
Risk Level: SAFENO_CODE
Full Analysis
- [SAFE]: The skill consists entirely of instructional markdown providing a workflow for legal analysis. It does not contain any executable scripts, binaries, or automated shell commands.
- [DATA_EXPOSURE]: The instructions direct the agent to use reputable public legal resources such as gesetze-im-internet.de and dejure.org for norm verification. No patterns for accessing sensitive local files or exfiltrating data to unknown external servers were found.
- [PROMPT_INJECTION]: The skill includes instructions to avoid relying on model training data for legal citations, requiring live verification instead. This is a reliability constraint and does not attempt to bypass agent safety filters or override system instructions.
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to process external documents like administrative acts and protocols. While this presents an ingestion surface, the skill lacks dangerous capabilities such as file system writing or command execution that would allow an attacker to exploit this surface.
Audit Metadata