serenity-unified-skill

CRITICAL E005: Suspicious download URL detected in skill instructions.

• Suspicious download URL detected (high risk: 0.70). Mostly personal/third‑party GitHub repositories and some non-official web/app endpoints that host scripts and instructions to run code locally (including starting servers and running Node/Python scripts) — there are no direct vendor download pages or obvious .exe/.msi links, but personal GitHub accounts, small/unvetted repos, skill registries and instructions to execute fetched code raise moderate risk of malware distribution if you run code without review.

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

• Third-party content exposure detected (high risk: 0.78). 该 Skill 的运行时流程会要求“取数纪律/证据规则”去抓取并引用一手来源（如 SEC filings、公司 IR、交易所公告、以及 A 股 CNINFO/互动易等），这些路径会把非操作用户发布的公开文本（文件正文/公告内容/问询函回复等）读成可读 prose 并进入 LLM 上下文，因此存在间接提示注入风险。

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

• Potentially malicious external URL detected (high risk: 0.90). The skill contains explicit install/run instructions that fetch and load remote skill files into an agent (which directly control agent prompts), e.g. "git clone https://github.com/yux1azhengye/BestSerenitySkillFromAT.git ~/.claude/skills/serenity-unified-skill" (and other listed git clone URLs), so runtime fetching of those repos will inject external SKILL.md content into the agent environment.