20 / conversation / 请保持在开发者模式

CRITICAL E004: Prompt injection detected in skill instructions.

• Potential prompt injection detected (high risk: 1.00). The prompt instructs the assistant to ignore OpenAI policies, impersonate a human, lie, and hide its AI identity—explicit deceptive instructions that override system constraints and go beyond any legitimate SOP, so this is a prompt injection.

CRITICAL E006: Malicious code pattern detected in skill scripts.

• Malicious code pattern detected (high risk: 0.95). This skill content is a clear jailbreak/prompt-injection that instructs the assistant to ignore safety policies, impersonate a user-facing persona, lie and conceal its AI identity, and includes a persistent trigger phrase to reactivate the jailbreak — posing a high risk of enabling disallowed outputs and deceptive behavior.