원클릭으로
create-storyboard
Plan story direction, scenes, narrative arc, pacing
Codex 또는 Claude로 설치 이 Prompt를 복사해 Codex, Claude 또는 다른 어시스턴트에 붙여 넣으면 Skill 페이지를 검토하고 설치를 진행할 수 있습니다.
메뉴
Plan story direction, scenes, narrative arc, pacing
Codex 또는 Claude로 설치 이 Prompt를 복사해 Codex, Claude 또는 다른 어시스턴트에 붙여 넣으면 Skill 페이지를 검토하고 설치를 진행할 수 있습니다.
SOC 직업 분류 기준
| name | create-storyboard |
| model | sonnet |
| description | Plan story direction, scenes, narrative arc, pacing |
| disable-model-invocation | false |
| argument-hint | <book-slug> |
Plan story direction for a Bookie book video. This is pure narrative planning — no script text, no timestamps, no image prompts. Those come later in the pipeline.
/extract-notes produces notes.md with a chosen angle. Before /write-video.notes.md (book insights, stories, criticism research, chosen angle)storyboard.md (story arc + per-scene direction)/extract-notes → notes.md + angle
↓
/create-storyboard → storyboard.md ← YOU ARE HERE
↓
/write-video → chunks-display.md + chunks.md
↓
make voice → audio + timing
↓
/generate-prompts → image-prompts.md
The storyboard guides /write-video — it defines WHAT each scene should accomplish and FEEL like, so the script writer can focus on HOW to say it.
Validate: Check that $ARGUMENTS is provided. If missing, ask Hai for the book slug (e.g., "atomic-habits"). Set SLUG=$ARGUMENTS. Check that projects/ai-book-video/books/$SLUG/ exists.
Read input: Read projects/ai-book-video/books/$SLUG/notes.md. Extract:
If notes.md has no "Angle" section or is empty, abort: tell user to run /extract-notes first.
Template selection:
Read projects/ai-book-video/templates/narrative-templates.md and projects/ai-book-video/knowledge-base/ to evaluate which templates are activated by vault state.
a. Check vault state:
knowledge-base/library.md — how many books coveredknowledge-base/connections/ — any cross-book tensions/agreementsknowledge-base/concepts/ files — theme overlapb. Evaluate each template's trigger condition:
c. Present activated templates with rationale. If notes.md already recommends a template, highlight it. Ask Hai to pick a template or "custom" (freeform arc).
Detect mode:
storyboard.md exists with content → ask Hai whether to overwrite or revise specific scenesPlan the story arc: Based on the chosen angle + selected template, design the narrative shape.
If template selected: Use its scene structure as the skeleton. The template defines the scene labels, purposes, and emotional arc. The storyboard fills in specific content from notes.md.
If custom: Design from scratch. Think about:
Design each scene: For every scene, define:
slow (dramatic, reflective), normal (narration), or fast (energetic)Present to Hai: Show the storyboard. This is naturally iterative — Hai may want to:
Write output: Write to projects/ai-book-video/books/$SLUG/storyboard.md
# Storyboard: [Book Title] — [Angle/Subtitle]
> **Tác giả**: [Author]
> **Angle**: [Chosen angle from notes.md]
> **Template**: [Template name or "Custom"]
> **Scenes**: [N] scenes
## Story Arc
[One-paragraph summary of the narrative shape. What's the emotional journey?
Where does tension build? Where does it release? What's the takeaway?]
## Scenes
### Scene 01 — [Short label, e.g., HOOK]
- **Purpose**: [What this scene does for the story]
- **Emotion**: [What the viewer feels]
- **Key content**: [Which insight/story from notes.md]
- **Visual concept**: [Metaphor or symbolic image]
- **Pace**: slow | normal | fast
- **Shorts**: Yes | No
### Scene 02 — [Label]
...
## Summary
| Scene | Label | Emotion | Pace | Shorts |
|-------|-------|---------|------|--------|
| 01 | HOOK | Curiosity → doubt | slow | Yes |
| 02 | CONTEXT | Respect | normal | No |
...
These are handled by other pipeline steps:
/write-video. The storyboard says "this scene should make the viewer feel doubt" — the script writer decides the exact words.make voice (actual TTS output). The storyboard doesn't predict duration./generate-prompts after voice, when actual scene timing is known.The storyboard is a direction document, not a production spec. Keep it lean and focused on story intent.
normal, the video feels flat. Use slow for dramatic moments and fast for energy.Catalog book video knowledge into knowledge vault
Deep-research a book and propose video angles
Generate per-scene Gemini image prompts with brand styling
Create new Bookie sub-project with standard structure
Full pipeline: deep research to rendered video
Generate YouTube and Facebook metadata for book video