devops

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

• Third-party content exposure detected (high risk: 1.00). The skill's browser-rendering and crawler examples (references/browser-rendering.md and references/cloudflare-workers-advanced.md) explicitly navigate to and scrape public websites (e.g., page.goto('https://news.ycombinator.com') and page.goto(message.body.url')) and feed that HTML into an AI extractor, meaning untrusted third-party content is ingested and can directly influence subsequent tool actions (extraction/enqueuing/AI decisions).

MEDIUM W012: Unverifiable external dependency detected (runtime URL that controls agent).

• Potentially malicious external URL detected (high risk: 0.90). The browser-rendering examples perform a runtime fetch of external page content and inject it into the model prompt (e.g., await page.goto('https://news.ycombinator.com') followed by passing page.content() to ai.run), so https://news.ycombinator.com is used at runtime to control the AI prompt.