audit-website

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

• Third-party content exposure detected (high risk: 0.90). SKILL.md's "Сбор данных" step explicitly instructs the agent to run web_fetch URL and browser action=open targetUrl=URL (and snapshot the DOM) to fetch and analyze arbitrary public websites, meaning the agent will read untrusted third‑party content which can materially influence audit decisions.