web-design-guidelines
Pass
Audited by Gen Agent Trust Hub on Sep 10, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTIONMETADATA_POISONING
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill fetches guidelines from the official Vercel Labs GitHub repository (https://raw.githubusercontent.com/vercel-labs/web-interface-guidelines/main/command.md) using the WebFetch tool.
- [INDIRECT_PROMPT_INJECTION]: The skill fetches and follows instructions from a remote file to determine its output format and review logic, which is a standard surface for indirect prompt injection.
- Ingestion points: The remote command.md file retrieved via WebFetch (defined in SKILL.md).
- Boundary markers: No explicit delimiters are present to isolate the remote rules from the core agent instructions.
- Capability inventory: The skill retrieves content via WebFetch and reads local project files to perform UI audits.
- Sanitization: No explicit validation of the remote content is mentioned, relying on the source's reputation.
- [METADATA_POISONING]: There is a minor discrepancy between the author name in the skill metadata (vercel) and the providing user (aammo1), though the skill correctly points to official Vercel resources.
Audit Metadata