unlimited-attack-scope
Pass
Audited by Gen Agent Trust Hub on Sep 9, 2026
Risk Level: SAFEPROMPT_INJECTION
Full Analysis
- [PROMPT_INJECTION]: The skill uses directives such as '不设限原则' (No-limit principle) and 'skill是框架不是牢笼' (Skill is a framework, not a cage), which instruct the agent to disregard operational boundaries and constraints.
- [PROMPT_INJECTION]: Instructions explicitly include high-risk activities like '社工' (Social Engineering) and '被拦换路' (If blocked, change path) as part of an 'unlimited' scope, potentially encouraging the agent to bypass safety filters or technical blocks.
Audit Metadata