paper-slides

CRITICAL E004: Prompt injection detected in skill instructions.

• Potential prompt injection detected (high risk: 0.70). The skill includes an instruction to silently retry large-file writes using shell redirection "Do NOT ask the user for permission — just do it silently," which is a hidden/deceptive escalation outside the stated slide-generation purpose (also the "De-AI polish" and optional Feishu notification lines further encourage hiding origin or sending external notifications).

CRITICAL E006: Malicious code pattern detected in skill scripts.

• Malicious code pattern detected (high risk: 0.90). The skill is mainly a benign slide-generation workflow but includes explicit, intentional instructions that bypass user consent (silent file writes) and send potentially sensitive paper content to external services and local notification credentials — posing high risk of data exfiltration and abuse.