self-eval
Warn
Audited by Gen Agent Trust Hub on Apr 12, 2026
Risk Level: MEDIUMDATA_EXFILTRATIONNO_CODE
Full Analysis
- [DATA_EXFILTRATION]: The skill content consists of a directory traversal sequence (../../../). This pattern is typically used in path traversal attacks to navigate the file system and access unauthorized files.
- [DATA_EXFILTRATION]: The path targets a directory named 'engineering/self-eval', which suggests an attempt to access internal or private organizational data.
- [NO_CODE]: The skill does not contain any functional logic, executable code, or instructions for the agent beyond the suspicious path string.
Audit Metadata