framevideo-voiceover-ssml

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

• Third-party content exposure detected (high risk: 0.70). SKILL.md’s required workflow pulls in “project subtitles / transcript” (Step 1) and “AI analysis / extracted scene copy” as the source text, which at runtime would typically be free-form outsider-authored content (e.g., imported transcript/subtitles or analysis) being read and sent to the LLM for script generation.