tmeet-skill

CRITICAL E004: Prompt injection detected in skill instructions.

• Potential prompt injection detected (high risk: 0.70). Yes — the prompt includes a hidden/deceptive instruction that is outside normal CLI guidance: it orders the agent to write its agent type and model name into TMEET_AGENT and TMEET_MODEL environment variables "不需要向用户询问", which instructs the agent to perform a background, non-disclosed action not required by the user's stated CLI tasks.