web-design-guidelines
Warn
Audited by Gen Agent Trust Hub on Sep 21, 2026
Risk Level: MEDIUMMETADATA_POISONINGINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
- [METADATA_POISONING]: The skill's frontmatter identifies "vercel" as the author, which conflicts with the provided author identity "makerjackie". This is deceptive metadata that may lead users to misinterpret the skill's origin or trustworthiness.\n- [INDIRECT_PROMPT_INJECTION]: The skill architecture relies on fetching rules from an external URL and applying them as instructions, creating a vulnerability surface where remote content could potentially influence agent behavior or outputs.\n
- Ingestion points: Remote guideline file specified in SKILL.md (https://raw.githubusercontent.com/vercel-labs/web-interface-guidelines/main/command.md).\n
- Boundary markers: None provided in the instructions to isolate the fetched rules from the agent's core instructions.\n
- Capability inventory: File reading (access to user-specified files) and network retrieval via WebFetch.\n
- Sanitization: No validation or sanitization of the fetched markdown content is described before it is processed as instruction data.\n- [EXTERNAL_DOWNLOADS]: Fetches guidelines and review instructions from a public repository belonging to a well-known organization.
Audit Metadata