website-to-framevideo

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

• Third-party content exposure detected (high risk: 0.75). Outsider free text can enter the LLM context via Step 0’s runtime website capture: the workflow fetches the user-supplied URL and extracts readable page text (e.g., capture/extracted/visible-text.txt / asset-descriptions.md) which is then read in Steps 2–3 to draft the script/storyboard.

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

• Potentially malicious external URL detected (high risk: 0.80). The skill explicitly loads runtime JS from CDNs (e.g., GSAP at https://cdn.jsdelivr.net/npm/gsap@3.14.2/dist/gsap.min.js and Three.js at https://cdn.jsdelivr.net/npm/three@0.181.2/+esm) which will execute remote code in the composition/player environment and are described as required adapters/dependencies in the capabilities docs.