app-ui-surfaces-toolkit
Pass
Audited by Gen Agent Trust Hub on Aug 30, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The ChatInterface component in SKILL.md defines a pattern for rendering streamed text from an external API (/api/chat). This creates an ingestion surface for indirect prompt injection where malicious instructions in the API response could influence the agent or user.
- Ingestion points: The messages state in ChatInterface (SKILL.md) receives content from the /api/chat stream.
- Boundary markers: The implementation does not include specific markers to delimit untrusted content or warn the agent against executing instructions within it.
- Capability inventory: The skill provides access to application navigation (router.push in CommandPalette and OnboardingChecklist) and file management tools (FileUploadZone).
- Sanitization: The code renders content directly into the DOM using standard React bindings without additional sanitization layers to strip potential injection payloads.
Audit Metadata