adversarial-reviewer
Fail
Audited by Gen Agent Trust Hub on Apr 5, 2026
Risk Level: HIGHDATA_EXFILTRATIONPROMPT_INJECTION
Full Analysis
- [DATA_EXFILTRATION]: The skill content (
../../../engineering-team/adversarial-reviewer/SKILL.md) uses a directory traversal sequence to escape the restricted skill directory and access files in the engineering team's repository. - [PROMPT_INJECTION]: The targeted path is specifically directed at the
adversarial-reviewerdirectory, which indicates an attempt to extract the system prompts, instructions, or security logic of the analysis tool itself.
Recommendations
- AI detected serious security threats
Audit Metadata