uit-material-controls-app-web
Warn
Audited by Gen Agent Trust Hub on Nov 3, 2026
Risk Level: MEDIUMCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill instructs the agent to execute multiple commands, including an "unmasked typecheck" and a JavaScript module located at
../uit-screen-architecture-web/scripts/validate-app-structure.mjs. These actions involve running scripts from sibling directories that are outside the scope of the skill itself.\n- [PROMPT_INJECTION]: The skill contains a directive to "Do not read the validator source." This instruction explicitly forbids the agent from inspecting the code it is required to execute, which acts as a concealment pattern that bypasses the agent's ability to verify the safety or integrity of the external script.
Audit Metadata