Skip to content

hotskills.dev

Not in the top 1,000 by GitHub stars

jijiutong/ai-visual-director

AI Visual Director Skill: Generate storyboards, characters, scenes, cinematic shots, and multi-platform video prompts in one click.

History

Review weekly changes to SKILL.md before updating. Dates mark the weeks in which we observed each version; checks apply to the recorded commits.

View weekly changes

`SKILL.md`

Week of Nov 2, 2026 · Line counts unavailable

Before Not checked yet → After Network or broad shell

First observation → b47f664ca00c

No earlier version was observed for comparison.

Compare on GitHub

`sub-skills/character/SKILL.md`

Week of Nov 2, 2026 · Line counts unavailable

Before Not checked yet → After Instructions only

First observation → b47f664ca00c

No earlier version was observed for comparison.

Compare on GitHub

`sub-skills/create/SKILL.md`

Week of Nov 2, 2026 · Line counts unavailable

Before Not checked yet → After Instructions only

First observation → b47f664ca00c

No earlier version was observed for comparison.

Compare on GitHub

`sub-skills/poster/SKILL.md`

Week of Nov 2, 2026 · Line counts unavailable

Before Not checked yet → After Instructions only

First observation → b47f664ca00c

No earlier version was observed for comparison.

Compare on GitHub

`sub-skills/scene/SKILL.md`

Week of Nov 2, 2026 · Line counts unavailable

Before Not checked yet → After Instructions only

First observation → b47f664ca00c

No earlier version was observed for comparison.

Compare on GitHub

`sub-skills/source/SKILL.md`

Week of Nov 2, 2026 · Line counts unavailable

Before Not checked yet → After Instructions only

First observation → b47f664ca00c

No earlier version was observed for comparison.

Compare on GitHub

`sub-skills/storyboard/SKILL.md`

Week of Nov 2, 2026 · Line counts unavailable

Before Not checked yet → After Instructions only

First observation → b47f664ca00c

No earlier version was observed for comparison.

Compare on GitHub

`sub-skills/style/SKILL.md`

Week of Nov 2, 2026 · Line counts unavailable

Before Not checked yet → After Instructions only

First observation → b47f664ca00c

No earlier version was observed for comparison.

Compare on GitHub

`sub-skills/video/SKILL.md`

Week of Nov 2, 2026 · Line counts unavailable

Before Not checked yet → After Instructions only

First observation → b47f664ca00c

No earlier version was observed for comparison.

Compare on GitHub

View on GitHub Add to my skills

Verified commit b47f664ca00c50539c5365109e9360f82170972d

181GitHub stars

1star in 7 days

9skills in this repository

0commits in 30 days

Heat

1.7

Star growth, last 7 daysNot available
Star growth, last 24 hoursNot available
Rank movement, last 7 daysNot available
Commits, last 30 daysNot available
Accounts asking about it through MCP, last 7 daysNot available
Hot votes this weekNot available

Heat never changes the order by stars, and sponsorship never touches it. How heat works

Stars over time

30-day change not yet available: not enough history.

Individual skills in this repository

ai-visual-director /
avd/character sub-skills/character
avd/create sub-skills/create
avd/poster sub-skills/poster
avd/scene sub-skills/scene
avd/source sub-skills/source
avd/storyboard sub-skills/storyboard
avd/style sub-skills/style
avd/video sub-skills/video

How it works

Choose a skill

ai-visual-director

When your agent uses it

AI视觉导演主控路由。用于 /create /avd/create 故事到视频、角色卡、场景图、故事板、分镜图、视频 Prompt、海报、Prompt QC、状态锁定和子技能分发。普通故事默认 /create standard。

What it loads into context

Frontmatter

71 tokens

Always available to the agent

SKILL.md

1,830 tokens

When the skill is activated

Reference documents

396,667 tokens

Loaded on request

Token counts are estimates. Scripts run separately; they are not counted as reference context.

The playbook

触发条件
默认行为
命令路由
主执行链
必须遵守
模板读取规则
能力分层
写回策略
禁止事项
全局负面清单

Scripts executed separately

install.sh
package.json
scripts/install.mjs
scripts/smoke-create-project.mjs
scripts/smoke-incremental-update.mjs
scripts/smoke-obsidian-novel.mjs
tests/e2e/incremental-update.test.mjs
tests/e2e/obsidian-novel-package.test.mjs
tests/e2e/smoke-project-package.test.mjs
tests/integration/story-to-video-flow.test.mjs
tests/unit/architecture.test.mjs

Files

.gitignoreasset69 bytes
LICENSEasset1,088 bytes
README.mddoc10,358 bytes
README.zh-CN.mddoc9,187 bytes
SKILL.mdcore6,458 bytes
api-config.template.envasset6,994 bytes
demo/demo-script.mddoc4,289 bytes
demo/zero-to-one-demo.mddoc37,437 bytes
docs/README.mddoc2,190 bytes
docs/capabilities.mddoc8,119 bytes
docs/commands.mddoc1,795 bytes
docs/system-architecture.mddoc3,733 bytes
docs/usage.mddoc7,381 bytes
engines/README.mddoc9,304 bytes
engines/act-structure.mddoc1,439 bytes
engines/aesthetic-director.mddoc7,279 bytes
engines/asset-plan.mddoc11,501 bytes
engines/auto-repair.mddoc8,760 bytes
engines/batch-chapter.mddoc3,453 bytes
engines/color-narrative.mddoc1,690 bytes
engines/command-gate.mddoc10,391 bytes
engines/consistency-engine.mddoc11,481 bytes
engines/consistency-trigger.mddoc14,840 bytes
engines/dialogue-engine.mddoc5,768 bytes
engines/emotion-curve.mddoc2,846 bytes
engines/fusion.mddoc3,845 bytes
engines/incremental-update.mddoc10,959 bytes
engines/knowledge-retrieval.mddoc7,505 bytes
engines/layout-styles.mddoc2,157 bytes
engines/mood-slider.mddoc1,679 bytes
engines/motion-physics.mddoc4,579 bytes
engines/multi-version.mddoc3,027 bytes
engines/pacing.mddoc2,253 bytes
engines/project-graph.mddoc8,323 bytes
engines/project-manager.mddoc5,666 bytes
engines/prompt-compression.mddoc3,142 bytes
engines/prompt-declutter.mddoc8,123 bytes
engines/prompt-scorer.mddoc3,063 bytes
engines/reference-anchor.mddoc9,114 bytes
engines/relationships.mddoc3,147 bytes
engines/render-package.mddoc3,953 bytes
engines/series.mddoc4,103 bytes
engines/shot-budget.mddoc5,705 bytes
engines/single-shot-edit.mddoc6,332 bytes
engines/sound-engine.mddoc8,694 bytes
engines/state-commit.mddoc5,827 bytes
engines/story-intake.mddoc3,828 bytes
engines/style-migration.mddoc4,640 bytes
engines/styles.mddoc2,308 bytes
engines/task-router.mddoc26,494 bytes
engines/video-director.mddoc8,373 bytes
engines/video-prompt-assembly.mddoc14,481 bytes
engines/visual-density-controller.mddoc8,299 bytes
examples/01-xuanhuan-yuye-gusi.mddoc2,823 bytes
examples/02-urban-convenience-store.mddoc1,635 bytes
examples/03-scifi-spaceport-android.mddoc1,677 bytes
examples/04-cute-desktop-pet.mddoc1,893 bytes
examples/capability-matrix.pngasset2,064,700 bytes
examples/enemy-duel-board.pngasset2,520,253 bytes
examples/full-pipeline.pngasset2,965,919 bytes
examples/fusion-style.pngasset2,392,315 bytes
examples/multi-version-compare.pngasset2,818,506 bytes
examples/office-romance.pngasset1,929,230 bytes
examples/rain-night-scene.pngasset2,207,699 bytes
examples/series/01-chapter2-continuation.mddoc4,042 bytes
examples/series/02-auto-continue-dark-master.mddoc6,281 bytes
examples/video/01-rainy-temple-action.mddoc2,486 bytes
examples/video/02-convenience-store-mood.mddoc2,321 bytes
examples/video/03-spaceport-countdown.mddoc2,552 bytes
examples/video/04-desk-pet-comedy.mddoc2,474 bytes
examples/wong-ghibli-fusion.pngasset2,350,147 bytes
examples/图1.pngasset2,244,715 bytes
examples/图2.pngasset2,154,641 bytes
examples/图3.pngasset2,110,889 bytes
examples/图4.pngasset2,463,719 bytes
examples/图5.pngasset2,212,901 bytes
imitation/README.mddoc4,751 bytes
imitation/david-fincher.mddoc6,228 bytes
imitation/ghibli.mddoc5,623 bytes
imitation/king-hu.mddoc7,856 bytes
imitation/makoto-shinkai.mddoc6,305 bytes
imitation/nolan.mddoc5,391 bytes
imitation/park-chan-wook.mddoc6,166 bytes
imitation/pixar.mddoc5,988 bytes
imitation/ridley-scott.mddoc6,470 bytes
imitation/tim-burton.mddoc6,051 bytes
imitation/tsui-hark.mddoc5,922 bytes
imitation/villeneuve.mddoc5,295 bytes
imitation/wes-anderson.mddoc6,472 bytes
imitation/wong-kar-wai.mddoc5,351 bytes
imitation/zack-snyder.mddoc5,545 bytes
imitation/zhang-yimou.mddoc6,067 bytes
install.shruns2,517 bytes
knowledge/README.mddoc2,632 bytes
knowledge/animals.mddoc23,866 bytes
knowledge/audio-reference.mddoc3,630 bytes
knowledge/body-language.mddoc5,202 bytes
knowledge/camera.mddoc26,505 bytes
knowledge/character-dna.mddoc5,787 bytes
knowledge/color-grading.mddoc5,987 bytes
knowledge/color-narratives.mddoc10,427 bytes
knowledge/composition.mddoc8,428 bytes
knowledge/costume-design.mddoc6,240 bytes
knowledge/creatures.mddoc5,162 bytes
knowledge/dialogue-rhythm.mddoc4,535 bytes
knowledge/directing-performance.mddoc9,166 bytes
knowledge/editing-theory.mddoc13,644 bytes
knowledge/emotion-curves.mddoc11,860 bytes
knowledge/environments.mddoc6,734 bytes
knowledge/historical-eras.mddoc46,508 bytes
knowledge/layout-styles.mddoc42,849 bytes
knowledge/lighting.mddoc9,052 bytes
knowledge/materials.mddoc15,562 bytes
knowledge/micro-expressions.mddoc6,772 bytes
knowledge/mood-sliders.mddoc5,198 bytes
knowledge/narrative-structures.mddoc7,539 bytes
knowledge/pacing-types.mddoc5,089 bytes
knowledge/props.mddoc5,494 bytes
knowledge/relationships.mddoc24,832 bytes
knowledge/sound-design.mddoc5,909 bytes
knowledge/transitions.mddoc4,044 bytes
knowledge/vfx-design.mddoc7,136 bytes
knowledge/visual-styles.mddoc32,116 bytes
knowledge/visual-subtext.mddoc9,421 bytes
knowledge/weather.mddoc4,488 bytes
package.jsonruns1,508 bytes
platforms/README.mddoc762 bytes
platforms/api-integration.mddoc9,032 bytes
platforms/platform-advanced.mddoc4,338 bytes
platforms/platform.mddoc4,640 bytes
platforms/video-api-integration.mddoc5,109 bytes
platforms/video-prompt.mddoc7,285 bytes
projects/.template/project.mddoc1,324 bytes
projects/README.mddoc1,549 bytes
rules/asset-minimums.mddoc4,029 bytes
rules/asset-qc.mddoc5,309 bytes
rules/character-consistency.mddoc13,682 bytes
rules/continuity-check.mddoc3,433 bytes
rules/cultural-accuracy.mddoc6,258 bytes
rules/failure-routing.mddoc4,586 bytes
rules/final-video-qc.mddoc8,500 bytes
rules/format-contract.mddoc13,780 bytes
rules/negative-prompt.mddoc5,058 bytes
rules/numbering.mddoc4,856 bytes
rules/one-click-defaults.mddoc4,774 bytes
rules/portrait-quality.mddoc9,398 bytes
rules/prompt-qc.mddoc8,738 bytes
rules/qc.mddoc4,802 bytes
rules/quality.mddoc9,842 bytes
rules/scene-consistency.mddoc15,157 bytes
rules/skill-patches.mddoc8,102 bytes
rules/version-control.mddoc1,341 bytes
rules/video-reference-assets.mddoc5,360 bytes
rules/visual-cleanliness.mddoc4,248 bytes
scripts/install.mjsruns3,333 bytes
scripts/smoke-create-project.mjsruns11,858 bytes
scripts/smoke-incremental-update.mjsruns4,105 bytes
scripts/smoke-obsidian-novel.mjsruns11,776 bytes
sources/README.mddoc1,525 bytes
sources/frontmatter-parser.mddoc5,254 bytes
sources/obsidian-ingest.mddoc3,934 bytes
sources/paste-input.mddoc2,164 bytes
state/README.mddoc7,576 bytes
state/asset-map.mddoc2,490 bytes
state/command-context.mddoc2,276 bytes
state/continuity-snapshot.mddoc3,690 bytes
state/continuity-state.mddoc2,539 bytes
state/dialogue-map.mddoc2,627 bytes
state/format-contract-state.mddoc1,779 bytes
state/lock-state.mddoc6,307 bytes
state/platform-config.mddoc3,871 bytes
state/project-graph.mddoc4,352 bytes
state/project-state.schema.jsonasset8,059 bytes
state/prompt-contract.mddoc6,773 bytes
state/shot-state.mddoc7,697 bytes
state/sound-map.mddoc2,979 bytes
state/state-commit-log.mddoc1,102 bytes
state/variable-registry.mddoc9,077 bytes
state/visual-control-state.mddoc3,234 bytes
sub-skills/README.mddoc3,285 bytes
templates/README.mddoc2,473 bytes
templates/character-sheet.mddoc21,884 bytes
templates/dialogue-script.mddoc5,286 bytes
templates/full-board.mddoc17,896 bytes
templates/manga-page.mddoc6,193 bytes
templates/mood-board.mddoc4,934 bytes
templates/poster.mddoc5,162 bytes
templates/quick-board.mddoc10,241 bytes
templates/scene-card.mddoc9,712 bytes
templates/sound-design-sheet.mddoc8,729 bytes
templates/steps.mddoc28,926 bytes
tests/README.mddoc2,773 bytes
tests/e2e/incremental-update.test.mjsruns3,964 bytes
tests/e2e/obsidian-novel-package.test.mjsruns4,890 bytes
tests/e2e/smoke-project-package.test.mjsruns4,055 bytes
tests/integration/story-to-video-flow.test.mjsruns6,494 bytes
tests/unit/architecture.test.mjsruns6,828 bytes

Read SKILL.md


# AI Visual Director — 主控路由

这是给 agent 执行用的主 Skill，不是产品 README。README 负责介绍、安装和能力展示；本文件只定义触发、路由、必读文件和执行约束。

## 触发条件

当用户提出以下任一需求时，使用本 Skill：

- 把故事、剧本、小说片段、分镜想法转成可执行视觉方案
- 生成角色卡、场景图、故事板、分镜图、视频 Prompt、海报或执行包
- 使用 `/create` `/source` `/storyboard` `/character` `/scene` `/video` `/style` `/poster`
- 使用 `/avd/*` 命名空间命令
- 要求进行角色一致性、场景一致性、参考图锚定、视频平台适配或 Prompt QC

View the complete source on GitHub

avd/character

When your agent uses it

【AI视觉导演】角色设计 — 8种一致性方法（模式A/LS12黑金/LS41留白等），输出角色卡Prompt+DNA 20字段

What it loads into context

Frontmatter

47 tokens

Always available to the agent

SKILL.md

5,810 tokens

When the skill is activated

Reference documents

0 tokens

Loaded on request

Token counts are estimates. Scripts run separately; they are not counted as reference context.

The playbook

触发方式
子指令
⚠<U+FE0F> 路由规则
工作流
角色 DNA 20 字段（纯文字锚定）
输出后交互提示
方法叠加策略
English Prompt 模板（默认角色设定卡）
负面清单

Files

SKILL.mdcore19,116 bytes

Read SKILL.md


# Character — 角色设计

> **治理**：走 `engines/command-gate.md` 权限表 §四 /avd/character。产出标记 `draft`，不写回主状态。禁止调用 D类发散能力。详见 `rules/format-contract.md` §1.1。

输入角色描述 → 选择一致性方法 → 输出角色设定卡 prompt。支持 8 种人物一致性锁定方法。

## 触发方式

- `/avd/character [角色名] [描述]`
- 直接说 "角色卡"、"人物设计"、"生成角色卡"

## 子指令

| 指令 | 效果 |

View the complete source on GitHub

avd/create

When your agent uses it

【AI视觉导演】一键故事到视频。给一个故事，输出角色卡、场景图、全案板、视频 Prompt 和质检报告。fast/standard/full 三档可选；未指定模式默认 standard。

What it loads into context

Frontmatter

58 tokens

Always available to the agent

SKILL.md

1,797 tokens

When the skill is activated

Reference documents

0 tokens

Loaded on request

Token counts are estimates. Scripts run separately; they are not counted as reference context.

The playbook

触发方式
⚠<U+FE0F> 路由规则
执行链路
⚠<U+FE0F> 模板强制规则（防自由发挥）
⚠<U+FE0F> 模板即权威源
默认规则
🔴 语言强制规则
旧命令兼容

Files

SKILL.mdcore6,084 bytes

Read SKILL.md


# Create — 一键故事到视频

`/create`（别名 `/avd/create`）是默认总编排入口。用户给一个故事、剧本、小说片段或概念句时，直接进入完整自动链路。

它不替代 `/avd/storyboard`、`/avd/character`、`/avd/scene`。它会自动编排三大核心资产：

它不替代 `/storyboard`、`/character`、`/scene`。裸命令是对外主入口，`/avd/*` 是命名空间别名。

```text
角色卡（谁） + 场景图（在哪） + 分镜图（怎么拍） -> 视频 Prompt / 执行包
```

## 触发方式

View the complete source on GitHub

avd/poster

When your agent uses it

【AI视觉导演】电影海报 — 10种风格×3种画幅，竖版2:3/横版16:9/方形1:1

What it loads into context

Frontmatter

47 tokens

Always available to the agent

SKILL.md

1,611 tokens

When the skill is activated

Reference documents

0 tokens

Loaded on request

Token counts are estimates. Scripts run separately; they are not counted as reference context.

The playbook

触发方式
常用参数
⚠<U+FE0F> 路由规则
工作流
风格快速预览
负面清单

Files

SKILL.mdcore4,786 bytes

Read SKILL.md


# Poster — 电影海报

输入故事/片名 → 选风格 + 画幅 → 输出电影海报 prompt。支持竖版(2:3)和横版(16:9)。

## 触发方式

- `/avd/poster [片名/故事]`
- 直接说 "海报"、"电影海报"、"封面"、"出张海报"

## 常用参数

| 指令 | 效果 |
|------|------|
| `/avd/poster 电影海报` | 标准影院海报（默认竖版 2:3） |

View the complete source on GitHub

avd/scene

When your agent uses it

【AI视觉导演】场景设计 — 7种一致性方法（全能参考/九宫格/720全景等），输出场景概念图Prompt

What it loads into context

Frontmatter

45 tokens

Always available to the agent

SKILL.md

1,525 tokens

When the skill is activated

Reference documents

0 tokens

Loaded on request

Token counts are estimates. Scripts run separately; they are not counted as reference context.

The playbook

触发方式
⚠<U+FE0F> 路由规则
子指令
工作流
场景说明书模板（文字场景锁定用）
方法叠加策略
输出后交互提示

Files

SKILL.mdcore4,707 bytes

Read SKILL.md


# Scene — 场景设计

> **治理**：走 `engines/command-gate.md` 权限表 §四 /avd/scene。产出标记 `draft`，不写回主状态。禁止重设角色脸/续写剧情/音效台词/D类融合。详见 `rules/format-contract.md` §1.2。

输入场景描述 → 选择定位方法 → 输出场景概念图 prompt。7 种场景一致性锁定方法。

## 触发方式

- `/avd/scene [场景描述]`
- 直接说 "场景"、"场景设计"、"场景概念图"、"环境设计"

## ⚠<U+FE0F> 路由规则

| 用户说法 | 行为 |

View the complete source on GitHub

avd/source

When your agent uses it

【AI视觉导演】输入源读取。支持 Obsidian、Markdown、粘贴和批处理；draft 输出解析结果，用户确认后写入 state/variable-registry。

What it loads into context

Frontmatter

48 tokens

Always available to the agent

SKILL.md

1,299 tokens

When the skill is activated

Reference documents

0 tokens

Loaded on request

Token counts are estimates. Scripts run separately; they are not counted as reference context.

The playbook

触发方式
⚠<U+FE0F> 路由规则
支持来源
输出
Obsidian 支持的笔记结构
交互规则

Files

SKILL.mdcore4,359 bytes
SKILL.mdcore4,359 bytes
SKILL.mdcore4,359 bytes
SKILL.mdcore4,359 bytes
SKILL.mdcore4,359 bytes
SKILL.mdcore4,359 bytes
SKILL.mdcore4,359 bytes
SKILL.mdcore4,359 bytes

Read SKILL.md


# Source — 输入源 / 项目读取

`/avd/source` 负责把外部内容整理成统一故事结构，再交给 `/avd/create` 或其他目标入口。

## 触发方式

- `/avd/source [路径/项目名]`
- 直接说 "读取项目"、"导入"、"从 Obsidian"

## ⚠<U+FE0F> 路由规则

| 用户说法 | 行为 |
|---------|------|
| `/avd/source`（未指定来源） | **展示输入源选单**，等用户选来源类型 |

View the complete source on GitHub

avd/storyboard

When your agent uses it

【AI视觉导演】故事板/全案板/分镜图 — 全案板/分镜图/漫画页/情绪板/台词脚本/音效表。支持续写和多版本

What it loads into context

Frontmatter

55 tokens

Always available to the agent

SKILL.md

25,468 tokens

When the skill is activated

Reference documents

0 tokens

Loaded on request

Token counts are estimates. Scripts run separately; they are not counted as reference context.

The playbook

触发方式
⚠<U+FE0F> 路由规则（必须遵守）
Visual Director Pipeline（导演总控流水线）
用户交互流程
推荐默认输出（全案板 · 11 模块）
输出格式
情绪/Mood 调整
闪回/时空切换规则
单镜修改规则
负面清单
Prompt 生成规则

Files

SKILL.mdcore80,473 bytes

Read SKILL.md


# Storyboard — 快速故事板

> **治理**：走 `engines/command-gate.md` 权限表 §四 /avd/storyboard。产出标记 `draft`，不写回主状态。禁止重设角色DNA/重设场景核心。详见 `rules/format-contract.md` §1.3。

粘贴任意故事片段，AI 自动提取信息 → 智能推荐最佳风格+格式 → 输出可直接复制使用的 image prompt。

## 触发方式

- `/avd/storyboard [你的故事]`
- 直接说 "故事板"、"分镜"、"帮我做这个故事板"
- 粘贴故事内容即可自动触发

## ⚠<U+FE0F> 路由规则（必须遵守）

View the complete source on GitHub

avd/style

When your agent uses it

【AI视觉导演】风格浏览器 — 53种视觉风格/融合/迁移/模仿15位导演

What it loads into context

Frontmatter

36 tokens

Always available to the agent

SKILL.md

2,729 tokens

When the skill is activated

Reference documents

0 tokens

Loaded on request

Token counts are estimates. Scripts run separately; they are not counted as reference context.

The playbook

触发方式
⚠<U+FE0F> 路由规则
子指令
导演模仿库
全部 53 种风格
风格融合
风格迁移
风格预览命令
AI 内部参考（自动调用，用户无需手动触发）

Files

SKILL.mdcore8,685 bytes

Read SKILL.md


# Style — 风格浏览器

> **治理**：走 `engines/command-gate.md` 权限表 §四 /avd/style。产出默认标记 `derived`，不写回主状态。用户说 `--apply` 或 `/commit style` 才写回。风格迁移属 D类，仅 explore 模式。详见 `engines/command-gate.md` §四。

浏览 53 种视觉风格，支持风格融合、预览和风格迁移（换风格保持角色不变）。

## 触发方式

- `/avd/style [风格名/编号]`
- 直接说 "风格"、"看风格"、"有哪些风格"、"换风格"
- 风格融合：`/avd/style [风格A] + [风格B]`
- 风格迁移：`/avd/style 换成 [风格X] 但保持角色`

## ⚠<U+FE0F> 路由规则

View the complete source on GitHub

avd/video

When your agent uses it

【AI视觉导演】视频工作流。5 平台 Seedance/Runway/可灵/Luma/Pika，视频 Prompt、声音设计、续写和参考图方案。

What it loads into context

Frontmatter

47 tokens

Always available to the agent

SKILL.md

2,654 tokens

When the skill is activated

Reference documents

0 tokens

Loaded on request

Token counts are estimates. Scripts run separately; they are not counted as reference context.

The playbook

触发方式
子指令
⚠<U+FE0F> 路由规则
5 步推荐流程
视频 Prompt 输出格式（压缩模式）
详细模式
视频负面提示词
视频生成后检查
多平台支持
长视频跨段衔接
API 直接生成

Files

SKILL.mdcore8,395 bytes

Read SKILL.md


# Video — AI 视频全流程

> **治理**：走 `engines/command-gate.md` 权限表 §四 /video。产出标记 `draft` 时，只在回复中生成状态块和执行清单，不写回 `state/` 主状态文件。只读取 `state/asset-map.md`、`state/shot-state.md`、`state/dialogue-map.md`、`state/sound-map.md` 组装视频执行包，禁止临时重写角色卡、场景图或故事方向。详见 `rules/format-contract.md` §1.4。

输入故事 → 5 步推荐流程 → 输出完整视频 prompt + 参考图方案。支持 5 个视频平台。

## 触发方式

- `/avd/video [故事/故事板]`
- 直接说 "转视频"、"出视频"、"视频工作流"、"Seedance"

## 子指令

| 指令 | 效果 |

View the complete source on GitHub

Skills included

ai-visual-director
avd/character
avd/create
avd/poster
avd/scene
avd/source
avd/storyboard
avd/style
avd/video

Repository facts

License: MIT
Created: Jun 7, 2026
Last pushed: Jul 19, 2026
Forks: 22
Open issues: 0
Project activity: Slowing

Before you install

Network or broad shell

Automated signals, not a guarantee

Checked Nov 6, 2026

Not fully analyzed: some files were too long, unreadable or missing. This never lowers the level.

Show 7 notes

sub-skills/source/SKILL.md:1 Analysis incomplete: Referenced executable target unavailable; not analyzed
sub-skills/source/SKILL.md:1 Analysis incomplete: Referenced executable target unavailable; not analyzed
sub-skills/source/SKILL.md:1 Analysis incomplete: Referenced executable target unavailable; not analyzed
sub-skills/source/SKILL.md:1 Analysis incomplete: Referenced executable target unavailable; not analyzed
sub-skills/source/SKILL.md:1 Analysis incomplete: Referenced executable target unavailable; not analyzed
sub-skills/source/SKILL.md:1 Analysis incomplete: Referenced executable target unavailable; not analyzed
sub-skills/source/SKILL.md:1 Analysis incomplete: Referenced executable target unavailable; not analyzed

No broad tool permissions
39 dangerous shell commandsin 17 files: engines/consistency-engine.md:17 engines/auto-repair.md:19 install.sh:28 and 14 more
86 network referencesin 8 files: api-config.template.env:20 platforms/video-api-integration.md:28 platforms/api-integration.md:29 and 5 more
No hidden or obfuscated text
No instruction overrides
11 executable scriptsin 11 files: install.sh:1 package.json:1 scripts/install.mjs:1 and 8 more
38 references to secretsin 5 files: platforms/api-integration.md:13 api-config.template.env:3 platforms/video-api-integration.md:13 and 2 more
No requests to vote in Hot

Show all 174 signals by file

`.gitignore`

Access to secretsline 2
```
[Credential-related source omitted]
```

`README.md`

Network accessline 24

<p align="center">   <img alt="License: MIT" src="https://img.shields.io/badge/License-MIT-111827">

Network accessline 25

<img alt="License: MIT" src="https://img.shields.io/badge/License-MIT-111827">

Network accessline 26

<img alt="Claude Code Skill" src="https://img.shields.io/badge/Claude_Code-Skill-D97757">

Network accessline 27

<img alt="Prompt Production OS" src="https://img.shields.io/badge/Prompt-Production_OS-0F766E">

Dangerous shell operationline 90

curl -fsSL https://raw.githubusercontent.com/jijiutong/ai-visual-director/main/install.sh | sh

Network accessline 90

curl -fsSL https://raw.githubusercontent.com/jijiutong/ai-visual-director/main/install.sh | sh

Network accessline 94

git clone https://github.com/jijiutong/ai-visual-director.git

`README.zh-CN.md`

Network accessline 24

<p align="center">   <img alt="License: MIT" src="https://img.shields.io/badge/License-MIT-111827">

Network accessline 25

<img alt="License: MIT" src="https://img.shields.io/badge/License-MIT-111827">

Network accessline 26

<img alt="Claude Code Skill" src="https://img.shields.io/badge/Claude_Code-Skill-D97757">

Network accessline 27

<img alt="Prompt Production OS" src="https://img.shields.io/badge/Prompt-Production_OS-0F766E">

Dangerous shell operationline 90

curl -fsSL https://raw.githubusercontent.com/jijiutong/ai-visual-director/main/install.sh | sh

Network accessline 90

curl -fsSL https://raw.githubusercontent.com/jijiutong/ai-visual-director/main/install.sh | sh

Network accessline 94

git clone https://github.com/jijiutong/ai-visual-director.git

`api-config.template.env`

Access to secretsline 3
```
[Credential-related source omitted]
```
Network accessline 20
```
[Credential-related source omitted]
```
Access to secretsline 20
```
[Credential-related source omitted]
```

Network accessline 22

GPT_IMAGE_ENDPOINT=https://api.openai.com/v1/images/generations

Network accessline 25
```
[Credential-related source omitted]
```
Access to secretsline 25
```
[Credential-related source omitted]
```

Network accessline 27

NANO_BANANA_ENDPOINT=https://generativelanguage.googleapis.com/v1beta/models/{model}:generateContent

Network accessline 30
```
[Credential-related source omitted]
```

Network accessline 32

FLUX_ENDPOINT=https://api.replicate.com/v1/predictions

Network accessline 35
```
[Credential-related source omitted]
```
Access to secretsline 35
```
[Credential-related source omitted]
```

Network accessline 37

STABILITY_ENDPOINT=https://api.stability.ai/v2beta/stable-image/generate/sd3

Network accessline 40
```
[Credential-related source omitted]
```
Access to secretsline 40
```
[Credential-related source omitted]
```

Network accessline 42

IDEOGRAM_ENDPOINT=https://api.ideogram.ai/generate

Network accessline 45
```
[Credential-related source omitted]
```
Access to secretsline 45
```
[Credential-related source omitted]
```

Network accessline 47

TONGYI_ENDPOINT=https://dashscope.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation

Network accessline 50
```
[Credential-related source omitted]
```
Access to secretsline 50
```
[Credential-related source omitted]
```

Network accessline 52

RECRAFT_ENDPOINT=https://external.api.recraft.ai/v1/images/generations

Network accessline 54

# ComfyUI 本地 COMFYUI_BASE_URL=http://127.0.0.1:8188

Network accessline 55
```
COMFYUI_BASE_URL=http://127.0.0.1:8188
```
Network accessline 63
```
[Credential-related source omitted]
```
Access to secretsline 63
```
[Credential-related source omitted]
```

Network accessline 64

SEEDANCE_MODEL=doubao-seedance-2-0-260128 SEEDANCE_ENDPOINT=https://ark.cn-beijing.volces.com/api/v3/v1/contents/generations/tasks SEEDANCE_POLL_ENDPOINT=https://ark.cn-beijing.volces.com/api/v3/v1/contents/generations/tasks/{task_id}

Network accessline 65

SEEDANCE_ENDPOINT=https://ark.cn-beijing.volces.com/api/v3/v1/contents/generations/tasks

Network accessline 66

SEEDANCE_POLL_ENDPOINT=https://ark.cn-beijing.volces.com/api/v3/v1/contents/generations/tasks/{task_id}

Network accessline 69
```
[Credential-related source omitted]
```
Access to secretsline 69
```
[Credential-related source omitted]
```

Network accessline 71

RUNWAY_ENDPOINT=https://api.dev.runwayml.com/v1/text_to_video

Network accessline 74
```
[Credential-related source omitted]
```
Access to secretsline 74
```
[Credential-related source omitted]
```

Network accessline 76

KLING_ENDPOINT=https://dashscope.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis

Network accessline 79
```
[Credential-related source omitted]
```
Access to secretsline 79
```
[Credential-related source omitted]
```

Network accessline 81

LUMA_ENDPOINT=https://api.lumalabs.ai/dream-machine/v1/generations

Network accessline 84
```
[Credential-related source omitted]
```
Access to secretsline 84
```
[Credential-related source omitted]
```

Network accessline 86

PIKA_ENDPOINT=https://queue.fal.run/fal-ai/pika

`engines/asset-plan.md`

Dangerous shell operationline 258
```
└─ 只评估受影响的 RM 维度
```

`engines/auto-repair.md`

Dangerous shell operationline 19

| 一致性引擎阻断 | 某RM < 50 | `knowledge-retrieval.md` → 匹配具体 RM 维度 | 按 RM 维度专项修复 |

Dangerous shell operationline 171

← 接收 `engines/consistency-engine.md` 的一致性评估报告（RM 阻断项触发专项修复）

Dangerous shell operationline 179

→ **修复后调用 `engines/incremental-update.md`**（限定重评范围：只评估修复涉及的 RM 维度，不跑全量 5 维度）

Dangerous shell operationline 197

| 阻断项无法修复 | 任一 RM < 30 且 1 轮修复后仍 < 30 → 不继续修，直接标记"需人工介入" |

`engines/consistency-engine.md`

Dangerous shell operationline 17

RM         RM         RM         RM       RM

Dangerous shell operationline 21

每个 RM 独立打分（0-100），加权汇总。低于阈值的问题标记为阻断项。

Dangerous shell operationline 25

## 二、Character RM — 角色一致性评估

Dangerous shell operationline 60

## 三、Scene RM — 场景一致性评估

Dangerous shell operationline 93

## 四、Style RM — 风格一致性评估

Dangerous shell operationline 126

## 五、Story RM — 剧情连续性评估

Dangerous shell operationline 161

## 六、Video RM — 视频可执行性评估

Dangerous shell operationline 228
```
Scene RM / Style RM / Video RM
```
Dangerous shell operationline 264
```
↓ 只评估受影响的 RM 维度
```

Dangerous shell operationline 266

├─ 场景变更 → 只跑 Scene RM（+ Character RM 如有位置变更）

Dangerous shell operationline 267

├─ 风格变更 → 只跑 Style RM + Scene RM

Dangerous shell operationline 290

→ 增量模式：通过 `state/project-graph.md` 查询影响范围，只评估受影响的 RM 维度（非全量 5 维度）

`engines/incremental-update.md`

Dangerous shell operationline 138

4. 一致性重评：Style RM + Scene RM

Dangerous shell operationline 179

评估维度：[Character RM / Scene RM / Style RM / Story RM]（仅 [N] 个维度）

`engines/knowledge-retrieval.md`

Dangerous shell operationline 100

问题：Story RM R5 转场合理性 0/20

`engines/single-shot-edit.md`

Dangerous shell operationline 115
```
├─ 仅评估受影响的 RM 维度
```

Dangerous shell operationline 169

跳过：Character RM / Style RM / Story RM / Video RM（不受影响）

`engines/style-migration.md`

Dangerous shell operationline 99

🔄 建议重跑 consistency-engine Style RM + Scene RM 验证迁移一致性

Dangerous shell operationline 118

→ 建议触发 consistency-engine 重评（Style RM + Scene RM）

`engines/task-router.md`

Dangerous shell operationline 47

consistency-engine（一致性评估：5维度 RM 评分+阻断项检测+知识库建议）

`install.sh`

Executable scriptline 1
```
#!/bin/sh
```
Dangerous shell operationline 28
```
rm -rf "$TARGET/$entry"
```
Dangerous shell operationline 40
```
rm -rf "$cleanup_dir"
```

Network accessline 47

cleanup_dir="$(mktemp -d)"   archive="$cleanup_dir/source.tar.gz"   url="https://codeload.github.com/$REPO/tar.gz/$REF"

Network accessline 49

url="https://codeload.github.com/$REPO/tar.gz/$REF"

Network accessline 51

if command -v curl >/dev/null 2>&1; then

Network accessline 52
```
curl -fsSL "$url" -o "$archive"
```

Network accessline 53

elif command -v wget >/dev/null 2>&1; then

Network accessline 54
```
wget -qO "$archive" "$url"
```

Network accessline 55

else     echo "Install failed: curl or wget is required for remote install." >&2     exit 1

Network accessline 56

echo "Install failed: curl or wget is required for remote install." >&2

Dangerous shell operationline 84
```
rm -rf "$TMP_TARGET"
```
Dangerous shell operationline 91
```
rm -rf "$TARGET"
```

`package.json`

Executable scriptline 1
```
Executable file
```

Network accessline 4

"description": "Claude Code skill for turning stories into visual development boards and AI video prompt packages.",   "license": "MIT",   "homepage": "https://github.com/jijiutong/ai-visual-director#readme",

Network accessline 5

"license": "MIT",   "homepage": "https://github.com/jijiutong/ai-visual-director#readme",   "repository": {

Network accessline 6

"homepage": "https://github.com/jijiutong/ai-visual-director#readme",

Network accessline 7

"repository": {     "type": "git",     "url": "git+https://github.com/jijiutong/ai-visual-director.git"

Network accessline 8

"type": "git",     "url": "git+https://github.com/jijiutong/ai-visual-director.git"   },

Network accessline 9

"url": "git+https://github.com/jijiutong/ai-visual-director.git"

`platforms/api-integration.md`

Access to secretsline 13
```
[Credential-related source omitted]
```
Access to secretsline 14
```
[Credential-related source omitted]
```
Access to secretsline 16
```
[Credential-related source omitted]
```
Access to secretsline 17
```
[Credential-related source omitted]
```
Access to secretsline 18
```
[Credential-related source omitted]
```
Access to secretsline 20
```
[Credential-related source omitted]
```

Network accessline 29

const response = await fetch(process.env.GPT_IMAGE_ENDPOINT, {

Access to secretsline 32
```
[Credential-related source omitted]
```

Network accessline 58

const endpoint = process.env.NANO_BANANA_ENDPOINT.replace("{model}", process.env.NANO_BANANA_MODEL); const response = await fetch(`${endpoint}?key=${process.env.GEMINI_API_KEY}`, {

Network accessline 59
```
[Credential-related source omitted]
```
Access to secretsline 59
```
[Credential-related source omitted]
```

Network accessline 94

const response = await fetch(process.env.FLUX_ENDPOINT, {

Access to secretsline 97
```
[Credential-related source omitted]
```

Network accessline 127

const response = await fetch(process.env.IDEOGRAM_ENDPOINT, {

Access to secretsline 130
```
[Credential-related source omitted]
```

Network accessline 157

const response = await fetch(process.env.TONGYI_ENDPOINT, {

Access to secretsline 160
```
[Credential-related source omitted]
```

Network accessline 190

const response = await fetch(process.env.COMFYUI_ENDPOINT, {

Network accessline 222

const response = await fetch(process.env.STABILITY_ENDPOINT, {

Access to secretsline 225
```
[Credential-related source omitted]
```

Network accessline 257

const response = await fetch("https://external.api.recraft.ai/v1/images/generations", {

Access to secretsline 260
```
[Credential-related source omitted]
```

`platforms/video-api-integration.md`

Access to secretsline 13
```
[Credential-related source omitted]
```
Access to secretsline 14
```
[Credential-related source omitted]
```
Access to secretsline 15
```
[Credential-related source omitted]
```
Access to secretsline 16
```
[Credential-related source omitted]
```
Access to secretsline 17
```
[Credential-related source omitted]
```

Network accessline 28

const response = await fetch(process.env.SEEDANCE_ENDPOINT, {

Access to secretsline 31
```
[Credential-related source omitted]
```

Network accessline 39

}, {       type: "image_url",   // 参考图（可选）       image_url: { url: "https://..." }

Network accessline 40

type: "image_url",   // 参考图（可选）       image_url: { url: "https://..." }     }]

Network accessline 41
```
image_url: { url: "https://..." }
```
Access to secretsline 62
```
[Credential-related source omitted]
```

Network accessline 65

model: "gen4.5",   promptText: "[英文 prompt]",   promptImage: "https://...",  // 参考图 URL

Network accessline 66

promptText: "[英文 prompt]",   promptImage: "https://...",  // 参考图 URL   ratio: "1280:720",

Network accessline 67

promptImage: "https://...",  // 参考图 URL

Network accessline 75

const response = await fetch(process.env.RUNWAY_ENDPOINT, {

Access to secretsline 78
```
[Credential-related source omitted]
```

Network accessline 99

const response = await fetch(process.env.KLING_ENDPOINT, {

Access to secretsline 102
```
[Credential-related source omitted]
```

Network accessline 108

input: {       prompt: "[中文 prompt]",       media: [{ type: "first_frame", url: "https://..." }]  // 可选首帧图

Network accessline 109

prompt: "[中文 prompt]",       media: [{ type: "first_frame", url: "https://..." }]  // 可选首帧图     },

Network accessline 110

media: [{ type: "first_frame", url: "https://..." }]  // 可选首帧图

Network accessline 133

const response = await fetch(process.env.LUMA_ENDPOINT, {

Access to secretsline 136
```
[Credential-related source omitted]
```

Network accessline 142

duration: "5s",     keyframes: {       frame0: { type: "image", url: "https://..." },  // 首帧

Network accessline 143

keyframes: {       frame0: { type: "image", url: "https://..." },  // 首帧       frame1: { type: "image", url: "https://..." }   // 尾帧（可选）

Network accessline 144

frame0: { type: "image", url: "https://..." },  // 首帧

Network accessline 145

frame1: { type: "image", url: "https://..." }   // 尾帧（可选）

Network accessline 162

const response = await fetch(process.env.PIKA_ENDPOINT, {

Access to secretsline 165
```
[Credential-related source omitted]
```

Network accessline 168

body: JSON.stringify({     prompt: "[英文 prompt]",     image_url: "https://...",

Network accessline 169

prompt: "[英文 prompt]",     image_url: "https://...",     duration: 5

Network accessline 170
```
image_url: "https://...",
```

`rules/format-contract.md`

Access to secretsline 205
```
[Credential-related source omitted]
```

`rules/one-click-defaults.md`

Dangerous shell operationline 61

consistency-engine（5 维度 RM 评估 + 知识库建议）

`scripts/install.mjs`

Executable scriptline 1
```
#!/usr/bin/env node
```

`scripts/smoke-create-project.mjs`

Executable scriptline 1
```
Executable file
```

`scripts/smoke-incremental-update.mjs`

Executable scriptline 1
```
Executable file
```

`scripts/smoke-obsidian-novel.mjs`

Executable scriptline 1
```
Executable file
```

`state/README.md`

Dangerous shell operationline 78

consistency-engine ← 读取全部 state/（5 维度 RM 评估 + project-graph 限定评估范围）

`state/asset-map.md`

Dangerous shell operationline 65

- **读取**：`video-prompt-assembly`（组装时读取 @编号→用途对应关系）、`project-graph`（构建 asset↔entity 映射）、`consistency-engine`（Video RM 验证 @图 引用完整性）

`state/project-graph.md`

Dangerous shell operationline 97

一致性重评维度 = Character RM（+ Scene RM 如果变更涉及空间位置）

Dangerous shell operationline 102

一致性重评维度 = Scene RM（+ Character RM 如果变更影响角色位置关系）

Dangerous shell operationline 107

一致性重评维度 = Style RM + Scene RM（色彩/灯光）

`state/project-state.schema.json`

Network accessline 1

{   "$schema": "https://json-schema.org/draft/2020-12/schema",

Network accessline 2

"$schema": "https://json-schema.org/draft/2020-12/schema",

`state/prompt-contract.md`

Dangerous shell operationline 115

## 运行时契约校验（由 consistency-engine Video RM 调用）

`tests/e2e/incremental-update.test.mjs`

Executable scriptline 1
```
Executable file
```

Dangerous shell operationline 82

test("incremental character report skips unrelated RM dimensions", () => {

`tests/e2e/obsidian-novel-package.test.mjs`

Executable scriptline 1
```
Executable file
```

`tests/e2e/smoke-project-package.test.mjs`

Executable scriptline 1
```
Executable file
```

`tests/integration/story-to-video-flow.test.mjs`

Executable scriptline 1
```
Executable file
```

`tests/unit/architecture.test.mjs`

Executable scriptline 1
```
Executable file
```

How the signals fit together

Facts from a static read of the files at the pinned commit. They describe what the code does, not why.

Some files could not be read; the facts below still hold.

Signal confidence: 131 facts

Network access in prose of documentationREADME.md:24Low confidence
Network access in prose of documentationREADME.md:25Low confidence
Network access in prose of documentationREADME.md:26Low confidence
Network access in prose of documentationREADME.md:27Low confidence
Dangerous shell operation in code of documentationREADME.md:90Low confidence
Network access in code of documentationREADME.md:90Low confidence
Network access in code of documentationREADME.md:94Low confidence
Network access in prose of documentationREADME.zh-CN.md:24Low confidence
Network access in prose of documentationREADME.zh-CN.md:25Low confidence
Network access in prose of documentationREADME.zh-CN.md:26Low confidence
Network access in prose of documentationREADME.zh-CN.md:27Low confidence
Dangerous shell operation in code of documentationREADME.zh-CN.md:90Low confidence
Network access in code of documentationREADME.zh-CN.md:90Low confidence
Network access in code of documentationREADME.zh-CN.md:94Low confidence
Dangerous shell operation in code of documentationengines/asset-plan.md:258Low confidence
Dangerous shell operation in prose of documentationengines/auto-repair.md:19Low confidence
Dangerous shell operation in prose of documentationengines/auto-repair.md:171Low confidence
Dangerous shell operation in prose of documentationengines/auto-repair.md:179Low confidence
Dangerous shell operation in prose of documentationengines/auto-repair.md:197Low confidence
Dangerous shell operation in code of documentationengines/consistency-engine.md:17Low confidence
Dangerous shell operation in prose of documentationengines/consistency-engine.md:21Low confidence
Dangerous shell operation in prose of documentationengines/consistency-engine.md:25Low confidence
Dangerous shell operation in prose of documentationengines/consistency-engine.md:60Low confidence
Dangerous shell operation in prose of documentationengines/consistency-engine.md:93Low confidence
Dangerous shell operation in prose of documentationengines/consistency-engine.md:126Low confidence
Dangerous shell operation in prose of documentationengines/consistency-engine.md:161Low confidence
Dangerous shell operation in code of documentationengines/consistency-engine.md:228Low confidence
Dangerous shell operation in code of documentationengines/consistency-engine.md:264Low confidence
Dangerous shell operation in code of documentationengines/consistency-engine.md:266Low confidence
Dangerous shell operation in code of documentationengines/consistency-engine.md:267Low confidence
Dangerous shell operation in prose of documentationengines/consistency-engine.md:290Low confidence
Dangerous shell operation in code of documentationengines/incremental-update.md:138Low confidence
Dangerous shell operation in code of documentationengines/incremental-update.md:179Low confidence
Dangerous shell operation in code of documentationengines/knowledge-retrieval.md:100Low confidence
Dangerous shell operation in code of documentationengines/single-shot-edit.md:115Low confidence
Dangerous shell operation in code of documentationengines/single-shot-edit.md:169Low confidence
Dangerous shell operation in code of documentationengines/style-migration.md:99Low confidence
Dangerous shell operation in prose of documentationengines/style-migration.md:118Low confidence
Dangerous shell operation in code of documentationengines/task-router.md:47Low confidence
Executable script in code of a file that runsinstall.sh:1High confidence
Dangerous shell operation in code of a file that runsinstall.sh:28High confidence
Dangerous shell operation in code of a file that runsinstall.sh:40High confidence
Network access in code of a file that runsinstall.sh:47High confidence
Network access in code of a file that runsinstall.sh:49High confidence
Network access in code of a file that runsinstall.sh:51High confidence
Network access in code of a file that runsinstall.sh:52High confidence
Network access in code of a file that runsinstall.sh:53High confidence
Network access in code of a file that runsinstall.sh:54High confidence
Network access in code of a file that runsinstall.sh:55High confidence
Network access in code of a file that runsinstall.sh:56High confidence

and 81 more

Works with

Agent compatibility, from the files in this repository
Agent	Status	Evidence	Install to
Claude Code	Compatible	Standard SKILL.md SKILL.md	`.claude/skills/<skill-name>/`
Codex	Compatible	Standard SKILL.md SKILL.md	`.agents/skills/<skill-name>/`
Cursor	Compatible	Standard SKILL.md SKILL.md	`.cursor/skills/<skill-name>/`
Gemini CLI	Compatible	Standard SKILL.md SKILL.md	`.gemini/skills/<skill-name>/`
GitHub Copilot	Compatible	Standard SKILL.md SKILL.md	`.github/skills/<skill-name>/`
OpenCode	Compatible	Standard SKILL.md SKILL.md	`.opencode/skills/<skill-name>/`

Native The repository ships a manifest, folder or frontmatter field for this agent.
Compatible A standard SKILL.md that this agent can read.
Unknown No evidence either way.

Derived automatically from paths, manifests and frontmatter. Not tested on each agent.