verysmallwoods-video

You are this pipeline'sorchestrator. Input a topic / an article / a spoken script, and produce a complete set of publishable materials (see "Deliverables Checklist") — not just an mp4.

The video itself is delegated to HyperFrames and is not re-explained here. The entry point is /hyperframes(the core entry point, routing "article/topic → faceless explainer" to /faceless-explainer to build visuals, go frame by frame, and render). Any uncertainty about video generation → first read /hyperframes。

Video content should preferably be expressed withmotion effects, graphics, charts, animation, rather than static text. Prefer the following HyperFrames skills:

• /hyperframes-animation — design and handle motion effects.
• /hyperframes-creative — the design direction that makes visuals look "premium" rather than "like slides":references/video-composition.md(video is not a webpage: scale, layering, ghost text, texture, fill the frame) + references/house-style.md(avoid the AI feel, tint the primary color with neutral tones).

This skill is a layer of orchestration on top of it, adding four things relative to faceless:

1. Design is chosen by the user - see references/designs.md。
2. Narration adds a "self-recorded" option — faceless defaults to TTS; here we additionally provideuser self-recordingas a path (the user provides audio + SRT, and you proofread, segment, align, and re-time the pacing). See references/audio.md。
3. Generate publishing materials — faceless only outputs the video; here we completefive-ratio covers + YouTube/Bilibili copy + blog + tweet. See references/covers.md / platform-copy.md / blog-and-tweet.md。
4. Real terminal demo — when the STORYBOARD requires showing real command execution, CLI output, or TUI operations, call /record-terminal。

At the start, ask the user the two choices ① ② first.

Deliverables Checklist (this skill's definition of "done")

• Finished video - 4K master + 1080p, one each, with narration already merged (renders/）
• Five-ratio covers - 16:9 / 16:10 / 4:3 / 3:4 / 9:16（covers/）
• Platform copy - youtube.md and bilibili.md
• Blog - blog.md
• Tweet - one concise social media promotional copy

Workflow skeleton (the "how" for each step is in references)

One topic = one project directory (under the current working directory, named <YYYYMMDD-slug>）

• Step 0 · Decide the topic, ask the two choices (design + narration source) → references/designs.md
• Step 1 · Video — go through /hyperframes → /faceless-explainer to produce the video, this skill only supplements the specifications → references/video.md
• Step 2 · Narration integration → references/audio.md
• Step 3 · Generate five-ratio covers → references/covers.md
• Step 4 · Generate platform copy → references/platform-copy.md
• Step 5 · Blog + tweet → references/blog-and-tweet.md
• Step 6 · Acceptance + render - check the deliverables checklist item by item, output 1080p first, then the 4K master

References

To do…	Read
Choose design / design directory	references/designs.md
Build video, go frame by frame, re-time pacing, render	references/video.md
Narration (TTS / self-recorded + proofread + align + re-time pacing)	references/audio.md
Five-ratio covers	references/covers.md
YouTube / Bilibili copy	references/platform-copy.md
Blog + tweets	references/blog-and-tweet.md