design-auditor
Pass
Audited by Gen Agent Trust Hub on Oct 23, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes external HTML, CSS, and JSON files, which constitutes a potential ingestion surface for indirect prompt injection.
- Ingestion points: The scripts
ai_slop_detector.py,design_system_validator.py, andcolor_contrast_checker.pyread user-provided design artifacts. - Boundary markers: Not explicitly implemented; however, the skill uses deterministic Python scripts to parse the data rather than passing raw file contents directly to an LLM prompt.
- Capability inventory: The scripts are limited to reading input files and writing report JSON/text files to the local filesystem. There are no network or arbitrary command execution capabilities.
- Sanitization: Content is processed via regular expressions and standard JSON parsing.
- [SAFE]: Systematic review of the provided Python scripts (
design_scorer.py,ai_slop_detector.py,color_contrast_checker.py,design_system_validator.py) confirms they use only Python standard libraries. No network operations, credential harvesting, or obfuscation techniques were detected.
Audit Metadata