design-workflows
Pass
Audited by Gen Agent Trust Hub on Oct 31, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTIONEXTERNAL_DOWNLOADS
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill audits repository code and external brand assets, creating an indirect prompt injection surface. It explicitly instructs the agent to treat file contents as inert data and to ignore/flag any embedded instructions.
- [COMMAND_EXECUTION]: The skill executes internal Python scripts (contrast.py, validate_tokens.py, lint_hardcodes.py) for technical validation tasks. These scripts are bundled within the skill and perform specific, non-malicious functions.
- [EXTERNAL_DOWNLOADS]: The asset sourcing workflow downloads logos and icons from external GitHub repositories (Jxck-S/soaring-symbols and Jxck-S/airline-logos) via curl. The skill documentation correctly identifies these third-party assets as untrusted markup and provides instructions for XML validation.
Audit Metadata