ai-sdk-ui

Pass

Audited by Gen Agent Trust Hub on Mar 24, 2026

Risk Level: SAFE
Full Analysis
  • [SAFE]: The skill is a collection of React templates and documentation for the Vercel AI SDK. All code samples follow best practices for frontend development.
  • [DATA_EXPOSURE]: Templates like message-persistence.tsx demonstrate chat history persistence using browser localStorage. This is a standard functional requirement for chat applications and does not expose sensitive system credentials or files.
  • [EXTERNAL_DOWNLOADS]: The utility script scripts/check-versions.sh queries the official npm registry (npm view) to check for package updates. This is a common development tool and does not involve downloading or executing untrusted remote code.
  • [INDIRECT_PROMPT_INJECTION]: The skill implements UI surfaces that ingest and display data from AI responses. It correctly identifies the vulnerability surface and provides templates using react-markdown for structured rendering. Boundary markers are maintained through the SDK's message role structure (user, assistant, tool). No dangerous capabilities (like shell execution or file system writes) are present in the UI hooks.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 24, 2026, 10:23 AM
Security Audit — agent-trust-hub — ai-sdk-ui