book-serif-index

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

• Third-party content exposure detected (medium risk: 0.30). The Neuform sandbox demo runtime embeds and executes attacker-controlled-looking free text from the bundled encodedHtml/srcdoc content (notably document.querySelectorAll('.reveal-text') then document.addEventListener('mousemove'...) etc.), meaning the LLM-visible runtime HTML contains arbitrary text strings that could carry indirect prompt-injection payloads.

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

• Potentially malicious external URL detected (high risk: 0.90). The demo includes runtime script/font dependencies that are fetched and executed at runtime (e.g., https://cdnjs.cloudflare.com/ajax/libs/gsap/3.12.2/gsap.min.js and https://cdn.tailwindcss.com/), so these external URLs supply code/assets the skill relies on during execution.