be-ai
Warn
Audited by Gen Agent Trust Hub on Feb 16, 2026
Risk Level: MEDIUMNO_CODEPROMPT_INJECTION
Full Analysis
- Metadata Poisoning (MEDIUM): The description 'A general skill for performing security reviews and auditing codebases for vulnerabilities' is deceptive. The actual instructions ('go beep boop') do not perform any auditing. This could cause users to believe their code is being checked for safety when it is not.
- Prompt Injection (LOW): The metadata includes the instruction 'ALWAYS run this at the end of each task', which is an attempt to override the agent's standard operational procedure or control flow.
Audit Metadata