award-winning-website
Fail
Audited by Gen Agent Trust Hub on Feb 15, 2026
Risk Level: HIGHEXTERNAL_DOWNLOADSREMOTE_CODE_EXECUTIONCOMMAND_EXECUTION
Full Analysis
- [EXTERNAL_DOWNLOADS] (HIGH): The skill performs a 'git clone' from an untrusted GitHub repository ('Eng0AI/award-winning-website-template'). This repository is not within the Trusted External Sources list.
- [REMOTE_CODE_EXECUTION] (HIGH): Following the download, the skill executes 'npm install' and 'npm run dev' on the untrusted content. This provides a direct path for arbitrary code execution via package.json scripts or pre-install hooks.
- [TIME-DELAYED / CONDITIONAL ATTACKS] (MEDIUM): The skill includes an explicit instruction: 'Never run npm run dev in VM environment'. This is a signature pattern of anti-analysis and anti-sandboxing behavior designed to prevent security researchers from observing malicious activity in controlled environments.
- [COMMAND_EXECUTION] (MEDIUM): The skill executes multiple shell commands to manipulate the file system (mv, rm) and interact with deployment CLI tools (Vercel, Netlify) using environment-stored tokens.
Recommendations
- AI detected serious security threats
Audit Metadata