quark-download

HIGH W007: Insecure credential handling detected in skill instructions.

• Insecure credential handling detected (high risk: 1.00). The skill explicitly instructs asking the user for share extraction codes (passwords) and to re-validate/use them (e.g., via --stoken/passcode in detail/validate commands), which requires the agent to accept and embed secret values verbatim in commands/outputs, creating exfiltration risk.

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

• Third-party content exposure detected (high risk: 0.80). The skill explicitly fetches and ingests public, user-generated search results from PanSou (https://s.panhunt.com/api) and Quark share APIs (https://drive-pc.quark.cn/...) as shown in SKILL.md and scripts/quark_search.py, and those results (share URLs, stokens, file listings) are parsed and used to decide which links to validate, detail, and trigger save actions—so untrusted third-party content can materially influence tool behavior.