web-design-guidelines
Warn
Audited by Gen Agent Trust Hub on Sep 15, 2026
Risk Level: MEDIUMMETADATA_POISONINGINDIRECT_PROMPT_INJECTION
Full Analysis
- [METADATA_POISONING]: The skill's YAML frontmatter in
SKILL.mdattributes the author asvercel, which contradicts the actual author identityfcakyon. This deceptive metadata can mislead users into misjudging the trustworthiness and origin of the skill. - [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and analyze untrusted external data (user UI code files) which exposes it to potential indirect prompt injection attacks.
- Ingestion points: Specified UI files or file patterns passed via arguments analyzed by
SKILL.mdandreferences/web-interface-guidelines.md. - Boundary markers: Absent. The instructions do not define clear boundaries or isolation guidelines to prevent the agent from executing instructions embedded within the analyzed source files.
- Capability inventory: The skill reads files and outputs formatted compliance findings text. It does not invoke secondary code execution tools or perform unsafe network actions.
- Sanitization: Absent. No sanitization, escaping, or strict structural parsing is applied to the target source code before text inspection.
Audit Metadata