comic-drama-master

HIGH W007: Insecure credential handling detected in skill instructions.

• Insecure credential handling detected (high risk: 1.00). The skill explicitly requires the agent to display full TOS signed URLs (including signature/query parameters) and preserve URLs verbatim (e.g., in ), which forces the LLM to output sensitive signed tokens verbatim and therefore risks secret exfiltration.

CRITICAL E006: Malicious code pattern detected in skill scripts.

• Malicious code pattern detected (high risk: 0.90). The code enforces uploading user-generated files to a default remote TOS bucket and routinely reads/uses cloud credentials (AK/SK/VeFaaS tokens) to call external APIs and store artifacts, which creates a clear risk of intentional or inadvertent data exfiltration if the default bucket or provided credentials are controlled by a third party.

MEDIUM W011: Third-party content exposure detected (indirect prompt injection risk).

• Third-party content exposure detected (medium risk: 0.30). runtime path: references/screenplay-generator.md → step 1 调用 python scripts/web_search.py，该脚本会把“搜索摘要（含外部网页/资料的文字摘要）”作为文本输入返回给后续剧本生成流程，攻击者可通过提交故事创意影响搜索查询并注入毒化摘要内容。

MEDIUM W013: Attempt to modify system services in skill instructions.

• Attempt to modify system services in skill instructions detected (high risk: 1.00). 该 skill 明确要求在运行时检查并自动安装系统级软件 ffmpeg，示例命令中包含 sudo apt-get/yum 安装语句，直接引导获取提升权限并修改主机状态。