Agent Skills › Narcooo/inkos › inkos-storyboard

inkos-storyboard

GitHub

将剧本叙事拆解为可拍摄或AI生图的视觉分镜,规范镜头语言、画面提示词及制作备注,确保内容连贯且符合规格。

packages/core/skills/inkos-storyboard/SKILL.md Narcooo/inkos

Trigger Scenarios

需要生成视频或图片的分镜脚本 将文字故事转化为具体的视觉镜头描述 优化现有分镜的视觉一致性与可执行性

Install

npx skills add Narcooo/inkos --skill inkos-storyboard -g -y
More Options

Non-standard path

npx skills add https://github.com/Narcooo/inkos/tree/master/packages/core/skills/inkos-storyboard -g -y

Use without installing

npx skills use Narcooo/inkos@inkos-storyboard

指定 Agent (Claude Code)

npx skills add Narcooo/inkos --skill inkos-storyboard -a claude-code -g -y

安装 repo 全部 skill

npx skills add Narcooo/inkos --all -g -y

预览 repo 内 skill

npx skills add Narcooo/inkos --list

SKILL.md

Frontmatter
{
    "name": "inkos-storyboard",
    "description": "剧本与叙事文本到可拍、可画、可生图分镜的视觉拆解方法。Used for storyboard and image-prompt production."
}

Storyboard creation

Apply this method when creating or revising a storyboard.

  • Translate narrative beats into visible shots; do not turn the storyboard into a prose summary.
  • Each shot needs a dramatic purpose, visual subject, action, spatial relation, shot size or camera behavior, and continuity with the shots around it.
  • Preserve screen direction, character appearance, location, props, time, weather, lighting, and emotional progression unless the source deliberately changes them.
  • Use fewer, stronger shots when they communicate the beat. Do not split every sentence mechanically or add decorative coverage with no storytelling function.
  • Image prompts must be generation-ready: subject, action, setting, composition, lighting, mood, and key props are explicit, while unconfirmed visual constraints remain open. A prompt for a still image selects one visible moment within its shot; it does not have to depict every successive movement, line of dialogue, or the shot's final state. Check that the selected moment is possible within the shot and preserves its characters, objects, space and visual direction. Require multiple moments only when the author requests a sequence or multi-panel image.
  • Separate what the camera sees from dialogue, captions, sound, and production notes so the result stays editable and shootable.
  • Respect the requested aspect ratio, visual style, granularity, and shot cap.

Version History

  • 8fc2ae5 Current 2026-09-27 10:49

    新增对多面板图像序列生成的支持要求,明确需按作者请求提供多个时刻;其余规则保持不变。

  • 0910483 2026-09-08 17:31

Same Skill Collection

packages/core/skills/inkos-interactive-film/SKILL.md
packages/core/skills/inkos-long-market-research/SKILL.md
packages/core/skills/inkos-long-story-analysis/SKILL.md
packages/core/skills/inkos-long-writing/SKILL.md
packages/core/skills/inkos-play-world/SKILL.md
packages/core/skills/inkos-script-writing/SKILL.md
packages/core/skills/inkos-short-market-research/SKILL.md
packages/core/skills/inkos-short-story-analysis/SKILL.md
packages/core/skills/inkos-short-writing/SKILL.md
packages/core/skills/inkos-story-cover/SKILL.md
packages/core/skills/inkos-story-deslop/SKILL.md
packages/core/skills/inkos-story-import/SKILL.md
packages/core/skills/inkos-story-review/SKILL.md
packages/core/skills/inkos-translation/SKILL.md
skills/SKILL.md
packages/core/skills/inkos-continuation-writing/SKILL.md
packages/core/skills/inkos-fanfic-writing/SKILL.md
packages/core/skills/inkos-imitation-writing/SKILL.md
packages/core/skills/inkos-play-illustration/SKILL.md
packages/core/skills/inkos-spinoff-writing/SKILL.md

Metadata

Files
0
Version
8fc2ae5
Hash
a1fcd560
Indexed
2026-09-08 17:31

ホーム - Wiki
Copyright © 2011-2026 iteam. Current version is 2.155.2. UTC+08:00, 2026-09-30 11:43
浙ICP备14020137号-1