stitch-ui-designer
Pass
Audited by Gen Agent Trust Hub on Jun 13, 2026
Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill defines an autonomous execution workflow for UI generation tasks. It explicitly instructs the agent to perform actions such as
create_projectandgenerate_screen_from_textimmediately upon receiving a valid request, bypassing standard human-in-the-loop confirmation loops to prioritize execution speed. - [PROMPT_INJECTION]: The skill functions by ingesting untrusted user design requests and interpolating them into complex prompts for secondary generation tools, creating an attack surface for indirect prompt injection.
- Ingestion points: User-provided design descriptions (e.g., 'Design a dashboard for SaaS') processed in the workflow steps of
SKILL.md. - Boundary markers: The skill utilizes structured sections like
[Context],[Layout], and[Components]to organize the resulting prompt and maintain structural integrity. - Capability inventory: The skill is authorized to use tools for project management, automated UI generation, and screen data retrieval as listed in the
allowed-toolsfrontmatter andSKILL.mdinstructions. - Sanitization: The workflow does not explicitly describe any validation or sanitization of user-supplied text before its inclusion in downstream prompts.
Audit Metadata