skills/klotzkette/claude-fuer-deutsches-recht/softwarefehler-mangelhaftung-pruefen/Gen Agent Trust Hub
softwarefehler-mangelhaftung-pruefen
Pass
Audited by Gen Agent Trust Hub on Aug 23, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill implements a professional instructional framework for legal consultation without malicious intent.
- [PROMPT_INJECTION]: The instructions define a role for an IT law specialist but do not attempt to override system safety protocols, bypass filters, or reveal internal prompts.
- [DATA_EXFILTRATION]: No sensitive credentials, private file paths, or unauthorized network operations were found. The skill references well-known legal resources like dejure.org and openJur, which is standard for its intended purpose.
- [REMOTE_CODE_EXECUTION]: There are no patterns involving the download or execution of external scripts, and no software dependencies are requested.
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest user-provided data such as defect descriptions and system logs. While this represents a data ingestion surface, the skill lacks high-privilege capabilities (e.g., shell access or file system modification) that would make this surface exploitable.
Audit Metadata