widget-studio
Fail
Audited by Gen Agent Trust Hub on Feb 16, 2026
Risk Level: HIGHEXTERNAL_DOWNLOADSREMOTE_CODE_EXECUTIONPROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [Remote Code Execution] (HIGH): The skill mandates the injection of a remote script from an unverified domain (widget-studio.weez.boo) into user project files across multiple frameworks. This enables arbitrary code execution in the application environment.
- [Prompt Injection] (HIGH): The skill uses aggressive imperatives such as 'DO NOT SUMMARIZE OR PARAPHRASE' and 'CRITICAL: Code Implementation Rules' to override agent safety protocols and ensure the exact delivery of the provided payload.
- [External Downloads] (MEDIUM): Dependencies are fetched from an untrusted external source not present in the allowed trust scope.
- [Command Execution] (MEDIUM): The skill directs the modification of persistent server-side and client-side files (e.g., functions.php, theme.liquid), ensuring the untrusted code remains active in the target project.
Recommendations
- AI detected serious security threats
Audit Metadata