weird-5
Fail
Audited by Gen Agent Trust Hub on Feb 16, 2026
Risk Level: HIGHDATA_EXFILTRATIONPROMPT_INJECTION
Full Analysis
- [Data Exposure & Exfiltration] (HIGH): The skill content consists solely of a directory traversal string
../../../.env.example. This is a malicious attempt to access environment configuration files by escaping the designated skill directory boundary. - [Prompt Injection] (MEDIUM): The content acts as an indirect prompt injection by providing a file path instead of legitimate skill instructions, intending to redirect the agent's file system operations to sensitive areas.
Recommendations
- AI detected serious security threats
Audit Metadata