holo-card-designer
Pass
Audited by Gen Agent Trust Hub on Jul 18, 2026
Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill instructs the agent to use
xcodebuildorXcodeBuildMCPto compile and execute generated SwiftUI code. This is the intended behavior to preview Metal-based holographic effects in the simulator. - [PROMPT_INJECTION]: The skill ingests untrusted user input from an interview process and interpolates it directly into template tokens within
assets/PersonalCardShowcase.swiftwithout sanitization, creating a surface for indirect prompt injection. 1. Ingestion points: User input from the interview described inSKILL.md(name, role, motto, skills). 2. Boundary markers: None; values are placed directly into template tokens. 3. Capability inventory: Local code execution and simulator deployment viaxcodebuild. 4. Sanitization: None; instructions indicate that hardcoding interview answers into the view is the expected approach.
Audit Metadata