skills/sonvee/ai-skills/uni-helper/Gen Agent Trust Hub

uni-helper

Pass

Audited by Gen Agent Trust Hub on Mar 5, 2026

Risk Level: SAFEEXTERNAL_DOWNLOADS
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The documentation provides instructions to install multiple packages from the @uni-helper scope on NPM and use npm create uni@latest for project scaffolding.
  • [INDIRECT_PROMPT_INJECTION]:
  • Ingestion points: Reads local project file structures (e.g., src/pages, src/layouts) and Single File Component (SFC) content via Vite plugins (plugin-pages.md, plugin-layouts.md).
  • Boundary markers: Not present; the documentation does not specify delimiters or validation for the content being processed from the filesystem.
  • Capability inventory: The documented tools are capable of writing to the local filesystem (generating pages.json, manifest.json, and .d.ts files) and initiating network requests via the uni-network library.
  • Sanitization: Not present; no explicit sanitization of filesystem input before configuration generation is described.
  • [METADATA_POISONING]: There is a minor discrepancy between the skill author context ('Sonvee') and the author specified in the skill's YAML metadata ('FlippeDround').
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 5, 2026, 04:53 PM