YOCO PPT Video Generator

Problem

Creating narrated PPT videos usually involves repeatedly coordinating course context, slide templates, narration, avatars, and video layout. yoco-ppt-video collapses this workflow around YOCO’s /agent/** APIs, turning an existing deck or agent-generated content into a finished video pipeline: course resolution, template selection, slide parsing, script generation, TTS, subtitle/avatar options, and video polling.

How It Works

• Configuration and auth: prefer YOCO_BASE_URL and YOCO_SKILL_TOKEN; ~/.yoco_ppt_video_config.json is also supported. The runtime should target Node.js 22+ with global fetch and WebSocket.
• Course and template: match a named course through /agent/course/list, create one if missing, and select a template dynamically via /agent/ppt/templatePage instead of hardcoding a templateId.
• PPT content: generate outline, page content, and images for /agent/ppt/submitContent, or upload an existing deck and poll the returned pptId; watch pptName, slide limits, and placeholder imageAction values.
• Video options: infer avatar, voice, layout, and subtitle intent from natural language; avatars and voices are resolved independently, and no avatar is added by default.
• Script and synthesis: derive narration.json from /agent/ppt/pages, synthesize audio through /agent/tts/websocket, save page notes, then call /agent/ppt/submitVideo.

Boundaries and Notes

This skill depends on a YOCO account, Skill Token, and compatible runtime, and should not call worker protocols directly. If the user does not explicitly request a digital human, keep the no-avatar layout; local images must be resolved and uploaded before submission.