web-design-guidelines-vercel
Warn
Audited by Gen Agent Trust Hub on May 2, 2026
Risk Level: MEDIUMPROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
- [PROMPT_INJECTION]: Deceptive Metadata. The skill metadata in SKILL.md lists the author as 'vercel', which contradicts the actual provider 'oimiragieo'. This misrepresentation can mislead users regarding the source and safety of the skill.
- [PROMPT_INJECTION]: Indirect Prompt Injection Surface. The skill is configured to fetch instructions from a remote source and follow them without constraints.
- Ingestion points: SKILL.md (fetches from raw.githubusercontent.com/vercel-labs/web-interface-guidelines/main/command.md)
- Boundary markers: Absent; instructions explicitly direct the agent to 'follow it exactly as presented'.
- Capability inventory: Standard agent interaction and file access.
- Sanitization: None; the fetched content is used directly as instructions.
- [EXTERNAL_DOWNLOADS]: Fetches design guidelines and configuration from Vercel's official GitHub repository.
Audit Metadata