no-ai-design-slop
Pass
Audited by Gen Agent Trust Hub on Aug 28, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to process and analyze external, untrusted data to inform its design decisions. This creates a surface where instructions hidden within design documents could potentially influence agent behavior.
- Ingestion points: The
SKILL.mdfile instructs the agent to read "local instructions, tokens, DESIGN.md, screenshots, references, and existing components" (SKILL.md). - Capability inventory: The agent is authorized to modify website code, app interfaces, and mockups based on the analysis of these files.
- Boundary markers: The instructions do not define clear delimiters or "ignore instructions" tags when interpolating content from the
DESIGN.mdor component source code. - Sanitization: There is no explicit logic provided to sanitize or validate the content of the external design documents before processing.
- [EXTERNAL_DOWNLOADS]: The skill references several external websites and documentation repositories for design guidance.
- Evidence: The
REFERENCES.mdfile contains links to X.com, YouTube, GitHub, W3C standards, Apple Human Interface Guidelines, and Google Material Design documentation. - Context: These links are informational resources used to provide design context to the AI and do not involve the automated installation of packages or execution of remote scripts.
- Trusted Sources: All linked resources originate from well-known technology organizations, official documentation platforms, or the author's own GitHub repository.
Audit Metadata