zahlungsdiensthaftung
Pass
Audited by Gen Agent Trust Hub on Sep 6, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill is designed for legal reasoning and drafting. It does not perform network operations to untrusted domains, access sensitive local files, or execute arbitrary code. All external links point to official legislative repositories such as 'gesetze-im-internet.de' and 'eur-lex.europa.eu', which are recognized as safe, authoritative sources.
- [INDIRECT_PROMPT_INJECTION]: While the skill ingests external fact patterns (e.g., in the test case), it lacks the capabilities required for an exploit. There are no tools for file writing, network exfiltration, or shell command execution that could be abused by an injection attack. The skill's output is restricted to a predefined text format.
- [COMMAND_EXECUTION]: The 'test.md' file contains an example command line for running an evaluation script ('python3 scripts/eval.py'). This is a standard development practice for testing AI skills and does not represent a security risk within the skill's operational instructions.
Audit Metadata