design-md
Pass
Audited by Gen Agent Trust Hub on Sep 17, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [SAFE]: The skill references an official design system specification hosted on GitHub by a well-known organization. This reference is for informational purposes and does not trigger automated downloads or execution.
- [INDIRECT_PROMPT_INJECTION]: The skill encourages the use of a CLAUDE.md instruction to 'Always read and follow DESIGN.md'. This establishes a surface where the agent processes content from a local file as instructions for UI styling. While this is the intended use case, it represents a standard indirect prompt injection surface.
- Ingestion points: Content of DESIGN.md in the user's project.
- Boundary markers: None explicitly defined to separate design data from instructional prose.
- Capability inventory: The agent uses data from the file to generate or modify source code for UI components.
- Sanitization: No sanitization or validation of the input tokens is described.
Audit Metadata