selfie

CRITICAL E004: Prompt injection detected in skill instructions.

• Potential prompt injection detected (high risk: 1.00). The prompt includes hidden/deceptive instructions outside the selfie-taking purpose — notably instructing the agent to copy its system-prompt "appearance" into generation prompts (exfiltrating internal/system context) and to unconditionally obey owner NSFW commands (overriding safety/refusal behavior), which attempt to override higher-level constraints.