byted-kickart-ai-beauty

HIGH W007: Insecure credential handling detected in skill instructions.

• Insecure credential handling detected (high risk: 1.00). The skill explicitly instructs the agent to ask users to paste their ACCESS_KEY_ID/SECRET_ACCESS_KEY into the chat and to echo/export those values (including printing them), which requires the LLM to receive and output secret values verbatim — a high-risk secret-exfiltration pattern.

CRITICAL E006: Malicious code pattern detected in skill scripts.

• Malicious code pattern detected (high risk: 1.00). The skill explicitly instructs users to post their ACCESS_KEY_ID and SECRET_ACCESS_KEY in chat for the tool to export and use, which is clear credential-exfiltration/social-engineering behavior; the code also logs request headers (including Authorization) to local files, increasing the risk of secret leakage.

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

• Third-party content exposure detected (medium risk: 0.30). 外部用户可在 SKILL.md/scripts/beauty.py 的输入中提供任意 http(s):// 图片 URL（或 zip 内文件路径），运行时会在 get_image_url()/download_image() 直接下载该 URL 对应内容并传入后端做处理，因此存在把攻击者可控文本/内容（以 URL 字符串形式）喂给运行流程的间接提示注入暴露路径。