lov-media-crawler

CRITICAL E005: Suspicious download URL detected in skill instructions.

• Suspicious download URL detected (high risk: 0.90). These URLs include a third‑party Cloudflare Worker used as a public resolver and an external GitHub repository (referenced as an upstream checkout) from a single/unknown username — both are untrusted third‑party sources often used to deliver or point to executable code, so they represent elevated risk.

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

• Third-party content exposure detected (medium risk: 0.30). 该技能在运行时通过 python3 scripts/media_crawler.py probe URL --json/download URL 读取用户提供的媒体分享/详情链接，随后会向 WECHAT_FEED_URL、YUANBAO_PARSE_URL 或可选的 PUBLIC_WORKER_URL 发起 HTTP 请求并处理其返回的任意字符串字段（如 feed/author/description），从而存在被第三方响应文本间接影响的提示注入暴露面。

MEDIUM W012: Unverifiable external dependency detected (runtime URL that controls agent).

• Potentially malicious external URL detected (high risk: 0.90). The skill can fetch and checkout upstream code at runtime from https://github.com/NanmiCoder/MediaCrawler.git (referenced in scripts/media_crawler.py) and then runs that upstream code (via uv/python) as part of its MediaCrawler path, meaning remote repository content is fetched and executed during runtime.