video-transcript

HIGH W007: Insecure credential handling detected in skill instructions.

• Insecure credential handling detected (high risk: 1.00). The skill explicitly instructs asking for and embedding API keys in requests/commands (curl headers, Authorization: Bearer , and browser-act auth set <KEY>) and says the key is "stored in memory", which requires the LLM to handle or emit secret values verbatim.

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

• Third-party content exposure detected (medium risk: 0.30). SKILL.md’s runtime workflow ingests outsider-authored free text when it uses browser-act/network interception to fetch YouTube subtitle text (via timedtext JSON3 events[].segs[].utf8) and Douyin page/network-derived transcript content, which originate from the user-supplied video/caption data rather than first-party trusted documentation.