talking-head-hyperframes

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

• Third-party content exposure detected (medium risk: 0.30). 该 Skill 的运行时工作流通过读取项目内的用户提供文本文件（如 inputs/视频脚本.md、inputs/制作规格.md、inputs/素材计划.md、inputs/motion-contract.json、字幕 captions_*.json/srt/vtt）并在固定舞台校验/就绪门与证明过程中把这些内容解析进逻辑与证据，属于把外部作者自由文本直接喂给运行时流程的情形（尽管在项目文件层面而非在线抓取）。

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

• Potentially malicious external URL detected (high risk: 0.90). The skill includes runtime-loaded external scripts and registries (e.g., the GSAP CDN script at https://cdn.jsdelivr.net/npm/gsap@3.14.2/dist/gsap.min.js and the HyperFrames registry URL https://raw.githubusercontent.com/heygen-com/hyperframes/main/registry referenced in hyperframes.json) which are fetched and executed at runtime (headless capture/session and npm/bootstrap installs), so remote content can execute code and affect the agent's animation-map/proof outputs.