name: video-script-storyboard description: Generate complete video production packages including hooks, voiceover scripts, storyboard shot lists, on-screen text overlays, and thumbnail prompts. Use when users need to create short-form or long-form video content plans, generate AI video/image generation prompts for each shot, or produce structured storyboards for video production. Triggers on requests like "generate a video script", "create a storyboard", "plan a short video about X", or any video content creation task.
Video Script + Storyboard
Generate complete video production packages: hooks, scripts, storyboard shot lists, on-screen text, and thumbnail prompts — with AI-ready prompts for each shot.
Workflow
This skill follows a 5-step sequential workflow:
- Gather topic — Obtain the video subject/story from the user
- Confirm parameters — Clarify key production parameters
- Generate content — Produce the full video package
- User review & refine — Iterate based on feedback
- Final output — Export as Markdown document to
assets/
Step 1: Gather Topic
Ask the user for their video topic/story content. Accept any of:
- A simple topic (e.g., "Pikachu vs Bulbasaur")
- A story concept or narrative outline
- A product/brand to feature
- Reference URLs or images for inspiration
Step 2: Confirm Parameters
Confirm the following parameters through conversation. Skip any the user has already specified. Use defaults for unspecified optional parameters.
| Parameter | Required | Default | Description |
|---|---|---|---|
topic |
✅ | — | 视频主题/故事内容 |
targetAudience |
✅ | — | 目标受众 |
durationSeconds |
❌ | 8s | 单段视频最大时长 |
resolution |
❌ | 2K | 分辨率 |
aspect_ratio |
❌ | 竖屏 9:16 | 尺寸画幅比 (竖屏/横屏 + 比例) |
tone |
❌ | — | 音调描述 (活力/温暖/悬疑等) |
style |
❌ | — | 视觉风格 (see references/video-styles.md) |
roles |
❌ | — | 多角色描述 (照片/图片/文本描述) |
角色一致性:When roles are specified, generate detailed character appearance descriptions to ensure visual consistency across all shots and segments.
风格选择:If the user is unsure about style, consult references/video-styles.md to suggest appropriate options based on their content type.
Step 3: Generate Content
Generate the full video package. For each video segment, produce four content blocks:
3.1 Thumbnail Prompt (缩略图提示词)
- AI image generation prompt for the video cover/thumbnail
- Write in English for maximum AI tool compatibility
- Include composition, style, elements, and mood
3.2 Voiceover Script (旁白脚本)
- Spoken narration / voiceover content
- Match the specified tone and target audience
- Include hooks in the opening to grab attention
3.3 Shot List (分镜镜头表)
For each shot, specify:
- Time range (e.g., 00-03s)
- Shot type — Reference references/shot-types.md for shot sizes
- Camera movement — Reference references/shot-types.md for movement types
- Scene description — Reference references/scene-types.md for scene vocabulary
- Audio/SFX — Sound effects and music cues
- AI Prompt — English prompt for AI video/image generation tools
Shot timing rules:
- Each shot ≤
durationSecondsparameter - Total duration should match
durationTotal - Use transitions from references/shot-types.md
3.4 On-screen Text (屏幕文字)
- Timestamped text overlays (subtitles, titles, CTA)
- Include emoji and visual indicators where appropriate
- Final frame should include CTA (Subscribe, Like, Comment, etc.)
Output Format
The final output document MUST contain both formats as defined in templates/output-template.md:
- JSON 结构数据 — Complete structured JSON following the
video_info+segmentsschema. Place at the top of the document inside ajsoncode block. This enables downstream tools (AI video generators, automation scripts) to parse and consume the data programmatically. - Markdown 可读文档 — Human-readable Markdown with tables, blockquotes, and visual formatting. Place after the JSON block for easy review and collaboration.
Step 4: User Review & Refine
Present the generated content and invite the user to request adjustments:
- Modify shot content for specific time ranges
- Change style, tone, or camera angles
- Add/remove/edit voiceover lines or sound effects
- Adjust timing or segment structure
- Regenerate thumbnail prompts
Iterate until the user is satisfied.
Step 5: Final Output
Export the final approved content as a document containing both JSON and Markdown:
- Follow both output formats (JSON + Markdown) defined in templates/output-template.md
- JSON block first (complete structured data), then Markdown section (human-readable)
- Save to
assets/[topic-slug]-storyboard.md - Confirm the output path with the user
References
- references/shot-types.md — Shot sizes, camera movements, transitions
- references/video-styles.md — Visual styles, color tones, pacing guides
- references/scene-types.md — Scene categories and description vocabulary
- templates/output-template.md — Output format template (JSON + Markdown)