design-md-enforcer
Pass
Audited by Gen Agent Trust Hub on Sep 25, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data from external URLs to extract design tokens and CSS styles.
- Ingestion points: Raw HTML, inline styles, and CSS files fetched from user-provided URLs in 'Workflow 1'.
- Boundary markers: The 'DESIGN.md' template includes specific sections to separate 'Observed' data from 'Inferred' analysis, helping the agent distinguish between external input and its own reasoning.
- Capability inventory: The skill performs file-write operations to the '/design-md/' directory but does not exhibit network-write, shell execution, or dynamic code execution capabilities.
- Sanitization: No explicit sanitization of the fetched HTML/CSS is described, relying on the agent's parsing logic to extract values for markdown tables.
Audit Metadata