NYC

web-design-guidelines

Pass

Audited by Gen Agent Trust Hub on Feb 17, 2026

Risk Level: SAFEEXTERNAL_DOWNLOADSPROMPT_INJECTION
Full Analysis
  • EXTERNAL_DOWNLOADS (LOW): The skill retrieves guidelines from https://raw.githubusercontent.com/vercel-labs/web-interface-guidelines/main/command.md. Under the [TRUST-SCOPE-RULE], this finding is downgraded to LOW because the vercel-labs organization is a trusted GitHub entity. No scripts or executables are downloaded.
  • PROMPT_INJECTION (LOW): The skill exhibits an indirect prompt injection surface by dynamically ingesting instructions from a remote URL to guide its auditing logic.
  • Ingestion points: Remote markdown fetched via WebFetch and interpreted as guidelines.
  • Boundary markers: Absent; the skill does not explicitly define delimiters for the external content.
  • Capability inventory: The skill can read local files and perform network fetches, but does not have permission for shell execution or arbitrary file writes.
  • Sanitization: No sanitization of the remote content is mentioned.
  • Context: Because the source is trusted and the skill's capabilities are restricted to analysis and reporting, the severity remains LOW.
Audit Metadata
Risk Level
SAFE
Analyzed
Feb 17, 2026, 06:41 PM