trendwatch

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

• Third-party content exposure detected (high risk: 0.95). Outsider free text is ingested via Step 4 collection: TikTokApi/instaloader/yt-dlp extract competitor captions (desc) and top comments (text) from non-user-authored accounts, which are then used in Step 5 analysis and written into competitors/{handle}.md and LLM context.

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

• Potentially malicious external URL detected (high risk: 0.70). The skill explicitly performs runtime WebFetch/WebSearch of external pages (e.g. https://creativecenter.tiktok.com and competitor video URLs like https://www.tiktok.com/@handle/video/{ID}, plus App Store pages) and ingests those fetched contents to drive ideation and prompt generation, so external content fetched at runtime directly controls the agent's outputs.