frontend-design
Warn
Audited by Gen Agent Trust Hub on Oct 2, 2026
Risk Level: MEDIUMMETADATA_POISONINGINDIRECT_PROMPT_INJECTION
Full Analysis
- [METADATA_POISONING]: The skill README.md claims the author is 'Anthropic', contradicting the actual author 'aiagentskills'. This impersonation of a trusted organization is deceptive and likely intended to gain unearned trust. Additionally, the skill reports unverifiable popularity metrics such as '10.8k downloads' and '42002 stars' which are highly likely to be inflated to mislead users regarding the skill's legitimacy.
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to process arbitrary user-supplied requirements for frontend code generation, which presents a surface for indirect prompt injection attacks where malicious instructions could be embedded in the design brief.
- Ingestion points: User-supplied requirements for components, pages, or applications as defined in SKILL.md.
- Boundary markers: There are no delimited boundaries or safety instructions to treat user input as untrusted data.
- Capability inventory: The skill generates and implements functional code (HTML, CSS, JavaScript, React, Vue) which can perform complex client-side logic.
- Sanitization: No input sanitization or output validation mechanisms are specified in the instructions.
Audit Metadata