design-brief
Pass
Audited by Gen Agent Trust Hub on Oct 6, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted user input to generate file content.
- Ingestion points: The 'brief' input string defined in SKILL.md.
- Boundary markers: The instructions lack explicit delimiters or safety markers for the brief input when interpolating into the DESIGN.md and HTML outputs.
- Capability inventory: The skill uses the 'file_write' capability to create DESIGN.md and brief-preview.html.
- Sanitization: While specific design dimensions are mapped via a token resolution table, natural language descriptions are interpolated into markdown and HTML outputs without explicit sanitization or escaping instructions.
Audit Metadata