deterministic-design
Pass
Audited by Gen Agent Trust Hub on Sep 13, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADSMETADATA_POISONING
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes rendered UIs and screenshots to perform layout audits, which introduces a surface where malicious instructions could be embedded in the visual or code data being analyzed.
- Ingestion points: UI renders and screenshots as described in SKILL.md.
- Boundary markers: No explicit prompt delimiters or instructions to ignore embedded content are mentioned.
- Capability inventory: Utilizes vision-based judging and a layout-audit.js script for measurement.
- Sanitization: No specific sanitization or validation of the rendered content is described.
- [EXTERNAL_DOWNLOADS]: The documentation references external components, including sub-skills and a JavaScript measurement tool (layout-audit.js), hosted on a community GitHub repository (github.com/connerkward/deterministic-design-skill). These resources originate from an unverified community source that is not recognized as a trusted organization or well-known service.
- [METADATA_POISONING]: There is a discrepancy between the author name specified in the skill's YAML frontmatter (Conner K Ward) and the provided author context (sickn33). While not directly malicious, metadata inconsistency can be a indicator of deceptive intent.
Audit Metadata