website-to-hyperframes

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

• Third-party content exposure detected (medium risk: 0.30). In Step 1 “Capture & Understand”, the workflow runs npx hyperframes capture <URL> on a user-provided URL and then reads LLM-visible extracted artifacts like extracted/visible-text.txt (site text) and other extracted files (screenshots/tokens/asset descriptions), which are outsider-authored by the target website.

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

• Potentially malicious external URL detected (high risk: 0.90). The skill includes runtime HTML/JS that loads and executes remote resources (e.g., the dotlottie player and GSAP plugin) such as "https://cdn.jsdelivr.net/npm/@dotlottie/player-component@2.7.12/dist/dotlottie-player.js", "https://cdn.jsdelivr.net/npm/gsap@3.14.2/dist/MotionPathPlugin.min.js", and a Google Fonts import "https://fonts.googleapis.com/css2?family=Fraunces:opsz,wght@9..144,100..900&display=block", which will be fetched and executed in the composition runtime.