stitch-mcp-generate-screen-from-text
Pass
Audited by Gen Agent Trust Hub on Apr 2, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill facilitates the generation of high-fidelity UI screens through a specialized MCP tool (
generate_screen_from_text). The instructions are clear and focus on design visualization rather than code implementation.- [PROMPT_INJECTION]: The skill processes user-provided text descriptions to create UI layouts. To mitigate potential indirect prompt injection, it instructs the agent to transform raw user input into a specific, structured format ([Device] [Mode] [Screen Type]. [Style]. [Layout]. [Components].) rather than interpolating untrusted input directly into the tool call.- [DATA_EXPOSURE]: The usage examples contain a numeric project ID (3780309359108792857). This appears to be a public-facing or demo identifier for the Stitch service and does not expose sensitive user credentials or system secrets.- [COMMAND_EXECUTION]: While the skill requests 'Read' and 'Write' permissions, these are scoped to the agent's environment for managing design assets (e.g., retrieving HTML or screenshots) as described in the output handling section. No arbitrary or dangerous command execution was detected.
Audit Metadata