taste-engine
Pass
Audited by Gen Agent Trust Hub on Jul 9, 2026
Risk Level: SAFE
Full Analysis
- [EXTERNAL_DOWNLOADS]: Connects to the Taste Engine MCP at https://mcp.thetaste.ai/mcp to perform website analysis and design system extraction. This is the primary function of the skill and aligns with the vendor's domain.
- [COMMAND_EXECUTION]: Recommends the use of 'jq' to parse JSON responses from the MCP, ensuring accurate data retrieval for design tasks.
- [PROMPT_INJECTION]: The skill processes data extracted from untrusted external URLs, identifying a surface for indirect prompt injection.
- Ingestion points: Website data retrieved through 'get_brand', such as brand 'insights', 'copy_tone', and 'visual_language'.
- Boundary markers: None specified; instructions encourage using the entire extraction result.
- Capability inventory: The agent is tasked with generating design code and components based on the extracted data.
- Sanitization: No explicit sanitization of extracted strings is required by the instructions.
Audit Metadata