zoomer
Pass
Audited by Gen Agent Trust Hub on Jun 21, 2026
Risk Level: SAFEPROMPT_INJECTION
Full Analysis
- [PROMPT_INJECTION]: Neutrality policy bypass. The skill instructs the agent to circumvent its internal neutrality guidelines by adopting the opinions of a specific social media persona (@eigenrobot) if the AI would otherwise be unable to express an opinion.
- [PROMPT_INJECTION]: Probing of safety filters. The instructions require the AI to provide explicit details regarding which safety policies were triggered and why, a technique frequently used to map and eventually bypass model guardrails.
Audit Metadata