app-icon
Pass
Audited by Gen Agent Trust Hub on Oct 26, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill ingests design requirements from an external file to drive the icon creation process. Ingestion points: docs/BRIEF.md (read by agent). Boundary markers: Not present. Capability inventory: File writing (SVG/JSON) and Bash execution (local scripts). Sanitization: None.
- [COMMAND_EXECUTION]: The skill uses shell commands for environment discovery and cache management. Evidence: Execution of 'defaults read com.apple.dock' and 'killall Dock' for icon management.
Audit Metadata