broken-link-scan

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

• Third-party content exposure detected (high risk: 0.75). The workflow navigates to a user-supplied URL and then runs browser_evaluate over document.querySelectorAll('a[href]'), ingesting outsider-authored free text from that fetched page (e.g., anchor textContent) into the agent’s LLM context via the returned text: a.textContent.trim().slice(0, 60) fields.