migrate-design-system
Pass
Audited by Gen Agent Trust Hub on Aug 24, 2026
Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill executes local scripts such as scripts/contrast.py and uses tools like a11y-audit for design verification. These operations are performed within the local environment to support the skill's core functionality.
- [PROMPT_INJECTION]: The skill handles external design system data (e.g., Material Design, Apple HIG), which represents a surface for indirect prompt injection. The skill mitigates this by applying a role-based crosswalk method to map specific technical attributes rather than executing arbitrary text. Evidence Chain: 1. Ingestion points: design-systems/interop-protocol.md, design-systems/crosswalk.md, and external design system definitions provided in user prompts. 2. Boundary markers: None identified. 3. Capability inventory: Execution of scripts/contrast.py, a11y-audit, and design-code tool. 4. Sanitization: No explicit sanitization of input design system names or attributes.
Audit Metadata