skills/vercel-labs/json-render/react/Gen Agent Trust Hub

react

Pass

Audited by Gen Agent Trust Hub on Sep 15, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
  • Indirect Prompt Injection Surface: The skill describes a system for rendering JSON specifications into UI components, specifically mentioning its use for "rendering AI-generated specs" and streaming content from API endpoints via useUIStream. This architecture creates a surface where untrusted data could attempt to influence the agent's behavior or manipulate the user interface.
  • Ingestion Points: Data enters the system through the spec prop of the Renderer component and the useUIStream hook which fetches data from external endpoints.
  • Boundary Markers: The skill encourages the use of zod schemas within the defineCatalog function to enforce structure and validate component props, providing a layer of protection against malformed or unexpected data.
  • Capability Inventory: The framework supports an event system (emit), state mutation actions (setState, pushState), and $computed function calls, which allow for dynamic interactions based on the processed specification.
  • Sanitization: The framework leverages React's inherent protection against cross-site scripting (XSS) for text content and promotes strict schema validation for all component properties.
  • Dynamic Function Execution: The $computed expression allows the UI to call functions registered in the developer's registry. While the functions themselves are defined locally, their invocation and arguments are driven by the dynamic JSON specification.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 15, 2026, 06:57 AM
Security Audit — agent-trust-hub — react