design-taste-frontend
Pass
Audited by Gen Agent Trust Hub on Sep 15, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADSCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to process user-provided design briefs and reference signals to generate frontend code and manage project dependencies. This creates a surface where external input can influence the agent's code generation and execution. \n
- Ingestion points: The skill interprets a 'design read' based on user briefs and external reference signals (URLs or screenshots) mentioned in Section 0.A.\n
- Boundary markers: The skill does not provide explicit boundary markers or delimiters for user-supplied data.\n
- Capability inventory: The skill can generate complex React/Next.js code and generate shell commands for environment setup and package installation (e.g., 'npm install').\n
- Sanitization: No specific sanitization or validation routines for external content are described.\n- [EXTERNAL_DOWNLOADS]: The skill provides instructions for the agent to install numerous third-party UI libraries and icons from official registries and references assets from CDNs. \n
- Evidence: Appendices A and B list packages from Microsoft, Google, IBM, Shopify, and Atlassian.\n
- Note: These downloads target well-known technology companies and official service domains, representing standard professional development workflows.\n- [COMMAND_EXECUTION]: The skill instructs the agent to generate and execute shell commands for initializing design systems and installing project dependencies.\n
- Evidence: 'npx shadcn@latest init', 'npm install @material/web', and other install commands are listed in the reference material.\n
- Note: These commands are used solely for the primary purpose of setting up the requested development environment.
Audit Metadata