skills/wix/skills/wix-manage/Gen Agent Trust Hub

wix-manage

Warn

Audited by Gen Agent Trust Hub on May 20, 2026

Risk Level: MEDIUMCREDENTIALS_UNSAFEPROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
  • [CREDENTIALS_UNSAFE]: The documentation for integrating external calendars instructs the AI to collect sensitive user credentials in plain text. Specifically, in references/bookings/external-calendar-integration.md, the instructions for connecting Apple Calendar via credentials require the agent to ask the user for their Apple ID email and password, which poses a security risk if these are stored in conversational history or logs.
  • [PROMPT_INJECTION]: The skill contains several instances of imperative language designed to override the agent's default operational logic and bypass user clarification steps.
  • Evidence: In references/ecommerce/recommend-ecommerce-strategy.md, the AI is told to "Do NOT ask clarifying questions" and to "execute every step in order" without producing recommendations before calling specific APIs.
  • Evidence: In references/stores/create-product-from-image.md, the AI is instructed to use a specific text description as the "only source of truth" and is prohibited from re-analyzing images later.
  • [PROMPT_INJECTION]: The skill exhibits an attack surface for indirect prompt injection by processing untrusted user data and using it as parameters for administrative API calls.
  • Ingestion points: User-supplied images in references/stores/create-product-from-image.md and user-provided blog content in references/blog/how-to-create-blog-posts.md.
  • Boundary markers: Absent. The agent is directed to derive facts from these inputs without delimiters or warnings to ignore potential embedded instructions.
  • Capability inventory: The skill uses CallWixSiteAPI to perform high-privilege actions including product creation, site property modification, and metadata management.
  • Sanitization: There is no evidence of sanitization or validation of the content extracted from user inputs before its inclusion in API payloads.
  • [EXTERNAL_DOWNLOADS]: Multiple recipes include references to external images from well-known providers for use in product catalogs and blog posts.
  • Evidence: references/stores/bulk-create-products-with-options.md and references/media/upload-media-to-wix.md use URLs from images.unsplash.com.
Audit Metadata
Risk Level
MEDIUM
Analyzed
May 20, 2026, 10:43 AM
Security Audit — agent-trust-hub — wix-manage