tools-ui
Pass
Audited by Gen Agent Trust Hub on Mar 15, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill references external resources from
ui.inference.shandcloud.inference.sh, including a component registry JSON and image assets. - [COMMAND_EXECUTION]: The documentation provides commands for developers to execute via
npx, specificallynpx shadcnandnpx skills, to add UI components and related skills. This is a standard practice for the referenced UI framework. - [INDIRECT_PROMPT_INJECTION]: The UI components are designed to display data from tool calls, such as arguments (
args) and results (result). This constitutes an indirect prompt injection surface as it handles data that may originate from untrusted LLM outputs; however, this is inherent to the purpose of a tool-monitoring UI and requires sanitization at the application level.
Audit Metadata