web-search

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

• Third-party content exposure detected (high risk: 0.95). Outsider-authored free text from arbitrary public web pages is fetched at runtime (e.g., fetchUrlFromContext loads page.goto(url) and extracts document.body text/Readability content), then returned as result.content/parsed.content which the agent would place into LLM context.