modernisierung-559-bgb

Pass

Audited by Gen Agent Trust Hub on Oct 23, 2026

Risk Level: SAFE
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted user-provided text, such as modernization notices and building descriptions, which constitutes a potential injection surface. However, the skill lacks high-risk capabilities like file system access, network operations, or tool execution, making the surface generally acceptable.\n
  • Ingestion points: Input fields for 'Maßnahmenbeschreibung', 'Ankündigungsschreiben', and 'Persönliche Lage des Mieters' defined in SKILL.md.\n
  • Boundary markers: None; the instructions do not implement delimiters or safety warnings for embedded content.\n
  • Capability inventory: No dangerous tools or subprocess calls are defined in the skill logic.\n
  • Sanitization: No input validation or filtering is performed on the ingested text.
Audit Metadata
Risk Level
SAFE
Analyzed
Oct 23, 2026, 12:29 PM