Introduction

People creating AI videos often get stuck at the first step: they have a theme and assets, but lack a script ready for immediate production. Writing visuals, crafting prompts, and filling in voiceovers shot-by-shot is repetitive and time-consuming. The purpose of this Skill is to expand a “topic/keyword” into a structured script—providing storyboards, visual prompts, and voiceover copy all at once, thereby reducing the manual labor involved in the transition from idea to production.

What is this

One-sentence positioning: Based on a topic or keyword input by the user, it generates a complete video script, including storyboard descriptions, visual prompts, and voiceover copy.

Publishing & Maintenance: Published by aguo333, sourced from clawhub, located at @aguo333/ai-video-script, current version v1.0.0, license MIT-0. Upstream repository: https://clawhub.ai/aguo333/ai-video-script. It is also indexed on the SkillHub mirror page https://www.skillhub.cn/skills/clawhub_aguo333/ai-video-script, categorized under content-creation.

Target Audience: Short video creators, AI video producers, and content marketers. Trigger words: Video script, storyboard, AI video, short video copy, video planning.

A note on a discrepancy: The task description provided a clue named “AI口播脚本” (AI Spoken Script) and mentioned “verification of word count vs. duration matching,” with a rating of 99656 and 1160 downloads. These details are inconsistent with the official page/pre-fetched data—the official name is “AI Video Script Generator.” The documentation only describes adjusting duration based on the platform and does not explicitly state a “word count/duration verification” mechanism. The pre-fetched rating/downloads were 100000 / 32436. The following text relies on verifiable official materials; unsubstantiated claims are excluded from the main body.

Core Features

The capabilities described in the official SKILL.md are as follows, all of which can be verified on the upstream page.

Each script contains four parts:

  • Video Overview: Title, duration, style positioning, target audience.
  • Storyboard Table: Scenes, visual descriptions, camera movements, duration.
  • Visual Prompts: For AI painting/video tools, annotated for compatibility with mainstream tools like CogVideoX, Midjourney, SD, etc.
  • Voiceover Copy: Narration text, subtitle text, emotional annotation.

The generation flow consists of four steps:
1. Understand Requirements (Topic/Product/Story, Target Audience, Video Style, Duration Requirements)
2. Plan Structure (Opening hook, Middle core content, Closing call-to-action)
3. Generate Storyboard (Purpose of each shot, visual composition, camera movement)
4. Write Prompts & Copy (Consider AI tool characteristics, include style/lighting/composition, Chinese-English comparison; copy should be colloquial, rhythmic, and emotionally guided).

Prompts are provided in two standard formats:

  • CogVideoX: [Subject Action], [Scene Environment], [Camera Movement], [Style], [Quality]
  • Midjourney: [Subject], [Action/State], [Scene], [Lighting], [Style], --ar 16:9

Advanced Features: Batch generation (input topic + multiple angles/styles to output 3-5 different script versions), Style customization (four presets: Commercial / Story / Educational / Entertainment), Duration optimization (adjusting duration according to platform).

Installation & Activation

The installation command provided on the upstream page is as follows; simply copy and execute:

openclaw skills install @aguo333/ai-video-script

The page also provides a CLI entry (npx skills) and a Prompt method. Once installed, invoke it using trigger words (e.g., “Video Script”, “Storyboard”).

Regarding differences with Cursor / Codex CLI / Claude Code: The official materials do not separately explain the activation methods for each client. We will not speculate here; the openclaw command is the standard.

Typical Usage

The official documentation provides two reproducible examples; the structure below illustrates the output format using them.

Example 1: Product Introduction (30 seconds). Topic: “Smart Water Bottle Introduction,” Style: Tech-feel, Concise. Output structure: Overview (Duration 30s, Style, Goal) → Storyboard Table (4 scenes, each containing Visual, Shot, Prompt, Voiceover, Subtitle) → Complete Prompt List → Voiceover Script. For instance, Scene 1:

### Scene 1 (0-3s): Opening
- Visual: Close-up of water bottle, flowing light and shadow
- Shot: Slow push in
- Prompt: Smart water bottle on wooden table, soft natural lighting,
  product photography, minimalist background --ar 16:9
- Voiceover: Still troubled by forgetting to drink water?
- Subtitle: Smart Water Bottle

Example 2: Knowledge Popularization (60 seconds). Topic: “Why do people dream?”, Style: Scientific, Fun. Also expanded into 4 scenes, with corresponding visual prompts, voiceovers, and subtitles for each.

Reproduction Method: Simply provide “Topic + Duration + Style”, for example, “Product introduction, 30 seconds, tech feel”. If batch variants are needed, add “Generate 3-5 different angles/styles”.

Use Cases & Notes

Who it suits: Creators and marketers who need to quickly produce structured scripts and then hand them over to AI drawing/video tools for execution; especially suitable for scenarios requiring multi-version, multi-platform reuse of the same topic.

Duration optimization covers the following platform ranges (listed in official documentation for reference):

  • Douyin / Kuaishou: 15-60 seconds
  • Bilibili / YouTube: 3-10 minutes
  • WeChat Channels: 30 seconds - 3 minutes

Limitations & Pitfalls:

  • The platform durations and styles are suggested ranges provided by the documentation, not automatic verifications. The official SKILL.md does not describe a “word count vs. duration auto-matching verification” mechanism; strict alignment requires manually reviewing voiceover word count against duration based on speaking speed.
  • Pay attention to copyright regarding prompts and copy: prompts should be original, avoid infringing elements, and comply with platform content requirements and sensitive word regulations.
  • The value of the script depends on the specificity of the input topic; the more vague the topic, the more generic the storyboard and copy will be.

Conclusion

The value of this Skill lies in handing over the manual orchestration of “Topic to Storyboard” to a tool. It provides a storyboard table, prompts adapted for mainstream AI tools, and voiceover copy with emotional annotations all at once, while supporting batch variants and platform-based duration adjustments. It is the “skeleton generator” in the content creation workflow, not a finished video.

Official URL: https://clawhub.ai/aguo333/ai-video-script
SkillHub Mirror: https://www.skillhub.cn/skills/clawhub_aguo333/ai-video-script