geo

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

• Third-party content exposure detected (high risk: 0.90). The skill explicitly fetches and parses arbitrary public websites and platform pages (e.g., SKILL.md Phase 1 "Fetch homepage HTML (curl or WebFetch)" and scripts/fetch_page.py, citability_scorer.py, brand_scanner.py which crawl pages, Reddit, YouTube, Wikipedia, sitemaps, robots.txt and llms.txt), and the tool reads and uses that untrusted, user-generated third‑party content to compute scores and drive audit decisions.