openclaw-videotranslate-skill

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

• Third-party content exposure detected (high risk: 0.70). The skill ingests subtitle text from user-supplied subtitle files or extracted embedded subtitles (SRT/VTT) and then sends that free-form text into the LLM via the translation provider’s translate_batch calls, so if the subtitles originate from an outsider (e.g., downloaded/forwarded/embedded in a third-party video), that outsider-authored text becomes LLM context.