web-design-guidelines

Pass

Audited by Gen Agent Trust Hub on Apr 18, 2026

Risk Level: SAFEEXTERNAL_DOWNLOADSPROMPT_INJECTION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: Fetches design guidelines and output formatting rules from the Vercel Labs GitHub repository.
  • [PROMPT_INJECTION]: Metadata indicates the author is "vercel", which is inconsistent with the provided author context.
  • [PROMPT_INJECTION]: The skill presents an indirect prompt injection surface by dynamically ingesting and following instructions from a remote source.
  • Ingestion points: Fetches content from a raw GitHub URL in SKILL.md to define its operational logic.
  • Boundary markers: No delimiters or instructions are used to separate the fetched guidelines from the agent's core safety constraints.
  • Capability inventory: Includes file system read access for user-defined files and network read access via WebFetch.
  • Sanitization: The skill does not perform validation or escaping of the remote instructions before they are applied to the local context.
Audit Metadata
Risk Level
SAFE
Analyzed
Apr 18, 2026, 03:41 PM