street-portrait-artist
Pass
Audited by Gen Agent Trust Hub on Nov 1, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes user-provided images to extract visual features and generate new artwork. This creates a surface where an attacker could attempt to embed malicious instructions within image metadata or visual content.
- Ingestion points: User-supplied portrait references (noted in SKILL.md and README.md).
- Boundary markers: The skill includes explicit instructions to ignore inferred personality traits and stick to visible geometry (noted in references/impression-map.md).
- Capability inventory: Image generation and reference analysis tools (noted in references/production-workflow.md).
- Sanitization: Lacks explicit instructions to disregard text-based instructions that might be present in the source images.
Audit Metadata