Skip to main content
Agent skill

Storyboard

Turns user materials into a shot-by-shot storyboard with bilingual narration (Chinese + English VO, both required) and AI video prompts per shot. When the user gives a target path (especially .md), write the full storyboard there — do not only print it in chat. Style and duration are free. Use when the user asks for storyboard, shot list, narration, VO script, AI video prompts, image-to-video prompts, or a video plan from images/clips/copy.

Get started

Get the complete skill folder, including its scripts and reference files.

When to use this skill

Use Storyboard when an AI agent needs a reusable SKILL.md workflow for this job: Turns user materials into a shot-by-shot storyboard with bilingual narration (Chinese + English VO, both required) and AI video prompts per shot. When the user gives a target path (especially .md), write the full storyboard there — do not only print it in chat. Style and duration are free. Use when the user asks for storyboard, shot list, narration, VO script, AI video prompts, image-to-video prompts, or a video plan from images/clips/copy.

When not to use it

Skip Storyboard when the task is outside the creativity category, or when a more specific skill in this directory already covers the same workflow with clearer triggers.

How to install

Where will you use it?

Install scope
Claude Code folder setup
mkdir -p ~/.claude/skills/storyboard
# Save SKILL.md to ~/.claude/skills/storyboard/SKILL.md
  1. Claude Code personal install: place the skill and its bundled files at ~/.claude/skills/storyboard/SKILL.md. For a shared project, commit the folder at .claude/skills/storyboard/ instead.
  2. Current Claude Code detects SKILL.md changes in watched directories during the session. If you just created a top-level skills directory, run /reload-skills. Check that the skill is listed before testing its trigger.
  3. For Claude chat or Cowork in the Desktop app, manage uploaded skills in Customize > Skills; copying files into a local Claude Code folder is not a Desktop install. Other agents may use different install locations.

Full install guide for Claude, Cursor, and Codex →

Try a starter prompt

After installing the skill, paste this starter prompt into your agent to verify it triggers on a small task.

Starter prompt
Use the Storyboard skill for this task: Detect delivery target (file vs chat). Confirm any required inputs or files before making changes.

What supports this recommendation?

83/100 provisional content score. This measures documentation, practical detail and source signals. It is not a measured task success rate or a security certification.

Documentation
92/100
Practical detail
92/100
Source evidence
61/100
Source signals
70/100
Model compatibility evidence
  • claudeuntested

    No model-specific signal or recorded compatibility test was found.

  • chatgptuntested

    No model-specific signal or recorded compatibility test was found.

  • geminiuntested

    No model-specific signal or recorded compatibility test was found.

  • copilotuntested

    No model-specific signal or recorded compatibility test was found.

  • llamauntested

    No model-specific signal or recorded compatibility test was found.

  • perplexityuntested

    No model-specific signal or recorded compatibility test was found.

  • mistraluntested

    No model-specific signal or recorded compatibility test was found.

  • grokuntested

    No model-specific signal or recorded compatibility test was found.

Inferred compatibility is based on content signals. Check the app, its available tools and setup requirements before running a skill.

What this skill does

# Storyboard

From **user materials only**, produce:

1. **Narration** per shot — **Chinese and English** (both required unless user opts out) 2. **Video prompt** per shot — ready for i2v / t2v

Do not invent product facts, prices, logos, or claims not in the materials. Style and length are agent-chosen from the content (no fixed genre or duration).

## Inputs

Any mix of stills, clips, text brief, and constraints. Optional: a delivery file path.

If goal/audience is ambiguous, state one short assumption and proceed. Ask only when a missing fact would make the storyboard unusable.

## Delivery

| Mode | Behavior | |------|----------| | User gives a file path | Write the **full** storyboard markdown to that path. Chat: short confirm (path, runtime, shot count) only | | No path | Full storyboard in chat |

Treat a path as the **output file** when:

1. User says write to / save as / put in + path 2. Skill request is paired with **one** main `.md` / `.txt` document 3. Path is an empty or draft file clearly opened for this output

**Rules:**

- Always **read** the target file first if it exists. Non-empty body = text material; then **replace** with the full storyboard (keep useful facts in `## Materials` / Spine). - One doc + images → write to the doc; images are materials only. - Explicit output path + separate brief → write only to the output path. - Several docs, unclear target → prefer the path next to the skill invoke; if still ambiguous, ask once. - User names a directory only → create `storyboard.md` there.

## Workflow

1. Detect delivery target (file vs chat) 2. Inventory materials (read delivery file if present) 3. Infer Visual Style (once for the whole piece) 4. Design story spine (hook → develop → payoff / CTA) 5. Break into shots (one clear idea each; duration free) — **Shot 01 is the eyeball grab** (see Shot 01) 6. Write bilingual VO + video prompt per shot 7. Write Assembly (after Spine), then Shots 8. Consistency pass; deliver in required format

### Visual Style (once)

Medium, mood, color/light, camera language, motion energy, aspect (default **16:9**). Keep consistent across prompts unless a deliberate style shift is story.

### Spine patterns

- Product → desire/problem → reveal → benefit → close - Character → setup → turn → resolution - Trailer → hooks and escalating beats - Explain → claim → show → reinforce - Attention-scarce / fragmented viewing → **grab in Shot 01** → develop → payoff / CTA

Spine may note what Shot 01 uses to stop the scroll; that hook is **not** a summary of the whole piece.

### Shot rules

- One idea per shot; prefer a **visible cause of motion** - Reference still present → **i2v** (motion relative to that frame) - No frame → **t2v** (full self-contained scene) - Respect user aspect; default 16:9 - No invented logo text, legal claims, or competitor names

### Shot 01 (hook — grab attention)

Attention is fragmented: **Shot 01 must catch the eye**. It is an opening line that stops the scroll—not a summary of the video, and not necessarily the piece’s main thesis.

- Prefer a sharp cold open: tension, curiosity, contrast, surprise, or a vivid concrete moment - VO + picture work as a hook; soft mood builds, empty atmosphere, or early detail dumps fail here - The main argument / product reveal can land in later shots; Shot 01 only needs to pull the viewer in - If materials already give a hook line, keep it and amplify — do not invent facts

### Narration (bilingual default)

| Field | Use | |-------|-----| | **Chinese** | Spoken Chinese for TTS / record (or `(no VO)`) | | **English** | Spoken English for TTS / record (or `(no VO)`) |

- Same idea in both languages (natural spoken, not word-for-word). - Length vs duration: Chinese ~3–4 chars/sec; English ~2–3 words/sec. Tighten the side that overruns. - **Sync on edit:** if either VO changes, update the other VO and related fields (`Duration`, `Visual`, `Camera`, `Video prompt`) in the same edit. - User gives one language verbatim → keep it; write a natural equivalent for the other. - User asks for only one language → omit the other field.

### Video prompts

Self-contained for a generic model. Include: subject + action in time order, camera (frame + move), environment, light/color, motion intensity, continuity anchors.

- i2v: start with `Image-to-video from provided still: …` then motion (+ keep composition unless changing) - t2v: full scene + Visual Style cues - Prefer concrete verbs; avoid named living artists / unlicensed IP not in materials - Optional: `Avoid: flicker, morphing faces, extra limbs, watermark, text overlay` - **English** prompt body unless user requires Chinese

## Output format (required)

Section order is fixed: Materials → Visual Style → Spine → Assembly → Shots.

```markdown # Storyboard — [working title]

## Materials - [asset or note → role]

## Visual Style - Medium: - Mood: - Color / light: - Camera: - Motion energy: - Aspect: [e.g. 16:9]

## Spine [1–4 sentences: arc + rough total runtime]

## Assembly [shot order, total runtime, TTS/BGM/stitch notes if obvious]

## Shots

### Shot 01 — [short title — cold open / hook] - **Duration:** ~Xs - **Source:** [file / t2v / i2v from …] - **Visual:** [composition] - **Camera:** [framing + move] - **Chinese:** [spoken Chinese, or (no VO)] - **English:** [spoken English, or (no VO)] - **Video prompt:** ``` [full prompt — English preferred] ``` - **Avoid (optional):** …

### Shot 02 — … … ```

## Quality bar

- [ ] File path given → full markdown **saved there** (UTF-8); not chat-only - [ ] Section order: Spine → Assembly → Shots; every shot has Chinese + English + Video prompt (or user opt-out) - [ ] VO languages sense-aligned; durations match speak length - [ ] **Shot 01** grabs attention as a cold open (not a thesis summary; no slow-burn open) - [ ] Facts trace to materials; style coherent; i2v points at the right still - [ ] No video/image generation APIs unless user separately asks - [ ] No invented output path when user gave none → chat delivery only

## CLI

Copy-paste commands: [cli/storyboard.md](../../../cli/storyboard.md)

Intended uses

  • Detect delivery target (file vs chat)
  • Inventory materials (read delivery file if present)
  • Infer Visual Style (once for the whole piece)
  • Design story spine (hook → develop → payoff / CTA)
  • Break into shots (one clear idea each; duration free) — Shot 01 is the eyeball grab (see Shot 01)
  • Write bilingual VO + video prompt per shot
  • Write Assembly (after Spine), then Shots
  • Consistency pass; deliver in required format

Implementation notes

  • Review the source instructions and adapt inputs before running the workflow.

Try it, then tell us how it went

A copied prompt does not tell us whether the task worked. Report your result after checking the output. We collect your tool and outcome, never your files or prompt.