ampelsystem-tabellenausgabe
Pass
Audited by Gen Agent Trust Hub on Sep 3, 2026
Risk Level: SAFENO_CODE
Full Analysis
- [NO_CODE]: The skill consists entirely of instructional text and formatting rules for a language model. There are no scripts, binaries, or executable commands provided.
- [SAFE]: The skill defines a logical framework for evaluating the phrasing of employment references (Arbeitszeugnisse) based on German labor law (GewO, BGB). It instructs the agent on how to categorize specific phrases into a traffic light system (Red, Orange, Green) and how to format the output in tables.
- [DATA_EXPOSURE]: No hardcoded credentials, sensitive file paths, or network exfiltration patterns were detected. References to internal markdown files (e.g.,
references/quellenhygiene.md) are standard for modular skill documentation. - [PROMPT_INJECTION]: The instructions do not contain patterns for bypassing safety filters, ignoring previous instructions, or escalating privileges. The 'Geheimcode-Regeln' refers to the legal 'coded' language in German employment references, not malicious code obfuscation.
- [INDIRECT_PROMPT_INJECTION]: While the skill is designed to process external documents (employment references), it lacks any high-risk capabilities such as shell execution, network writes, or file modifications that could be abused via malicious content in the ingested data.
Audit Metadata