canary

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

• Third-party content exposure detected (medium risk: 0.30). The skill ingests outsider-authored free text via runtime HTTP responses by curl-fetching user-supplied ${URL} endpoints like /sitemap.xml and /robots.txt (Phase 3) and repeatedly fetching page content/health data with curl in the monitoring loop (Phase 4), where those responses can contain arbitrary text.