design-an-interface
Pass
Audited by Gen Agent Trust Hub on Jun 18, 2026
Risk Level: SAFENO_CODE
Full Analysis
- [SAFE]: The skill implements a structured workflow for architectural exploration, using sub-agents to generate radically different interface designs based on user requirements.
- [SAFE]: The skill ingests user input and interpolates it into sub-agent prompts, creating a surface for indirect prompt injection. However, this is categorized as safe because the skill only generates descriptive design content and lacks access to sensitive data or tools for execution or exfiltration. Evidence Chain: 1. Ingestion points: [module description] and [gathered requirements] in SKILL.md. 2. Boundary markers: Absent. 3. Capability inventory: Spawns sub-agents via Task tool for text generation; no network or file system access is indicated in the skill logic. 4. Sanitization: None.
- [NO_CODE]: The skill consists exclusively of instructional Markdown and contains no scripts or executable code.
Audit Metadata