suprasumo-design-start
Pass
Audited by Gen Agent Trust Hub on Nov 9, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes data from external websites and writes it to local markdown files, creating a potential attack surface.
- Ingestion points: Scraped content from external gallery URLs defined in
references/sources.md(Land-book, SceneAI, 60fps, Recent). - Boundary markers: Uses predefined markdown templates in
SKILL.mdforspec.mdandplan.md, though it lacks explicit instructions to the agent to disregard instructions embedded in the external content. - Capability inventory: Creates directories and writes markdown files (
specs/<slug>/spec.mdandplan.md) within the local project. - Sanitization: No explicit sanitization or escaping of external content is described before interpolation into files.
- [EXTERNAL_DOWNLOADS]: Fetches design reference metadata and gallery content from well-known design platforms including Land-book, SceneAI, 60fps, and Recent.
Audit Metadata