design-md
Pass
Audited by Gen Agent Trust Hub on Sep 26, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill utilizes
web_fetchto retrieve HTML source code and screenshots from URLs provided by the Stitch MCP server (e.g., htmlCode.downloadUrl). These assets are used to analyze the design tokens and visual style of a project.- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted HTML content from project screens to extract styles, which is a potential surface for indirect prompt injection. - Ingestion points: HTML code is ingested via
web_fetchduring the asset download phase. - Boundary markers: The instructions lack specific delimiters or guardrails to prevent the agent from following instructions embedded in the analyzed HTML.
- Capability inventory: The skill can perform file write operations and project-level metadata retrieval through the Stitch MCP interface.
- Sanitization: There is no explicit sanitization of the downloaded HTML content before it is processed.
Audit Metadata