skills/jezweb/claude-skills/ai-sdk-ui/Gen Agent Trust Hub

ai-sdk-ui

Pass

Audited by Gen Agent Trust Hub on Feb 17, 2026

Risk Level: SAFE
Full Analysis
  • [Indirect Prompt Injection] (SAFE): The skill provides UI components designed to ingest untrusted user input for processing by LLMs.
  • Ingestion points: React state managed inputs and textareas in various templates (e.g., templates/use-chat-basic.tsx, templates/use-object-streaming.tsx).
  • Boundary markers: Absent at the UI layer; delimiters are typically managed on the server-side within the system prompt.
  • Capability inventory: The templates utilize sendMessage, complete, and submit functions to interact with AI providers via the Vercel AI SDK.
  • Sanitization: Standard UI input handling is used; sanitization for LLM safety is expected at the backend API layer.
  • [Command Execution] (SAFE): The script scripts/check-versions.sh uses standard, read-only npm commands (npm list, npm view) to verify installed package versions. This is a common developer utility for environment auditing.
  • [External Downloads] (SAFE): The templates/package.json file lists standard, widely-used dependencies for React and AI SDK development. No suspicious or unverified packages were identified.
Audit Metadata
Risk Level
SAFE
Analyzed
Feb 17, 2026, 04:39 PM