ds-token-architecture
Pass
Audited by Gen Agent Trust Hub on Sep 14, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONNO_CODE
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted external data, such as user-pasted design token JSON data and source code stylesheets, creating a potential surface for indirect prompt injection. No specific validation or isolation boundary markers are defined for the token inputs. However, the system's capability inventory is strictly limited to standard design system analysis via Figma MCP tools without any shell execution or arbitrary network capabilities.
- [NO_CODE]: The entire skill package consists solely of markdown files, including procedural check descriptions, reference material, and report templates. No executable code files, scripts, or binary assets are bundled with the skill, eliminating common supply chain and direct execution risk vectors.
Audit Metadata