content-source-aggregator

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

• Third-party content exposure detected (high risk: 0.90). SKILL.md and the included scripts/fetch_all.py clearly fetch and parse untrusted, user-generated public content from many open sources (e.g., old.reddit.com/r/{sub}/hot.json, syndication.twitter.com, hacker-news/firebase API, YouTube RSS, 60s.viki.moe, B站 API, etc.) and ingest those results into a hotspot pipeline that other agents consume for topic selection and downstream actions, so third‑party text could indirectly inject instructions that influence behavior.