elements
Pass
Audited by Gen Agent Trust Hub on Oct 12, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSREMOTE_CODE_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [REMOTE_CODE_EXECUTION]: The skill instructs the agent to use
npx assistant-uiandnpx shadcnCLI tools for project initialization and component installation. These commands download and execute scripts from the npm registry, which is standard procedure for the assistant-ui ecosystem. - [EXTERNAL_DOWNLOADS]: UI component source code and configuration files are downloaded from the vendor's official registry at
r.assistant-ui.comduring the setup and installation process. - [INDIRECT_PROMPT_INJECTION]: The skill provides components that render untrusted data from LLM outputs, such as markdown and tool call results. This constitutes a potential attack surface for indirect prompt injection. * Ingestion points:
Thread,MarkdownText, andToolFallbackcomponents process model-generated content. * Boundary markers: Standard UI encapsulation is used to delimit rendered content. * Capability inventory: Components are focused on UI rendering and do not possess direct system or network capabilities. * Sanitization: Relies on internal React rendering and markdown parsing logic provided by the library.
Audit Metadata