The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill fetches design guidelines from a well-known service (Vercel Labs) via their public GitHub repository. This is a standard operation to ensure the audit rules are kept up to date.- [PROMPT_INJECTION]: The skill exhibits an indirect prompt injection surface as it ingests instructions from a remote URL to process user code. 1. Ingestion points: Fetches guidelines from a GitHub URL and reads user-provided UI files. 2. Boundary markers: None are defined in the skill logic. 3. Capability inventory: The skill has file-read and network-fetch capabilities. 4. Sanitization: There is no evidence of sanitization for the fetched content. However, the risk is mitigated by the fact that the remote source is a trusted organization.- [DATA_EXFILTRATION]: While the skill reads local UI code, it only uses it for the analysis process. There are no patterns indicating that the contents of these files are being exfiltrated to unauthorized external domains.

web-design-guidelines