Menu
视频素材分析与清点。解析 PDF 文档提取文本/图片/表格，组织原始素材目录，生成结构化的 inventory.yaml 素材清单。
Install with Codex or Claude Copy this prompt, paste it into Codex, Claude, or another assistant, and let it review the skill page and install it for you.
Based on SOC occupation classification
Remotion Shot 组件实现。将 storyboard 转化为 React 组件。包含关键的帧计算规则（Math.round 强制）、interpolate 安全规则（最小 1 帧）、1080p 视觉标准、动画编排模式、音视频同步架构、组件复用模式。这是技术密度最高的 skill。
Remotion 资产管道。调用 project_migrate tool 将 raws/ 素材迁移到 template-project/public/ 并生成 resourceMap.ts、constants.ts、timing.ts。Agent 只需 edit constants.ts 中的 COLORS 和 FONTS。
音频素材生成与获取。批量 Edge TTS 旁白生成（支持 storyboard pacing 字段驱动语速）、BGM/SFX 检索（BGM 节奏匹配 BPM 规则）、音频时长提取。包含 Edge voice 配置、速度调整规则、durations.json 格式规范（含 audio_visual_relation 说明）和关键的音频时序规则。
视觉素材获取与生成。根据 resources.yaml 定义，通过库存图片/视频搜索（Pexels/Pixabay）或 AI 生图（Caro LLM）获取视觉素材。包含搜索策略、视觉隐喻搜索（counterpoint 声画关系）、构图感知搜索（framing 字段）、prompt 工程指南、素材质量评估标准。
旁白脚本口语化处理与情绪节奏标注。
视频制作策划。四阶段交互式策划流程：Creative Memo → Resources → Script → Storyboard。包含完整的影视标准（剪辑七律、景别、机位、灯光）、盖曼讲故事法则（叙事引擎、冰山理论、叙事声音、灵感堆肥、人物推进、幽默感、对白法则）、视听语言系统（情绪曲线、景别/运镜/节奏/张力/声画关系/转场）、文档模板和审批流程。
| name | carocut-planner-analysis |
| description | 视频素材分析与清点。解析 PDF 文档提取文本/图片/表格，组织原始素材目录，生成结构化的 inventory.yaml 素材清单。 |
Extracts structured content from source materials (PDF, images, videos, audio) and creates an organized inventory for video production planning.
Transform user-provided raw materials into structured data that can drive production planning. This step creates the foundation for all subsequent workflow stages.
| Type | Extensions | Extraction Method |
|---|---|---|
| PDF documents | Text blocks, images, tables via material decompose tool | |
| Images | .png, .jpg, .jpeg, .webp | Copied to raws/images/existing/ |
| Videos | .mp4, .mov, .webm | Copied to raws/videos/ |
| Audio | .mp3, .wav, .m4a | Copied to raws/audio/ |
| Text | .txt, .md | Read and indexed |
| URL | http/https | Text, images, tables extracted via web crawling |
raws/
images/
existing/ # User-provided images (organized)
crawled/ # Images downloaded from URL
crawled_img_001.jpg
crawled_img_002.png
page_01_img_01.png # Images extracted from PDF
page_02_img_01.png
table_01.png # Table screenshots from PDF
data.json # Structured content from PDF or URL
content.txt # Plain text extraction
inventory.yaml # Complete material inventory
{
"stats": {
"total": 45,
"pages": 12,
"by_type": { "text": 30, "title": 5, "image": 7, "table": 3 }
},
"elements": [
{ "type": "title", "page": 1, "content": "Introduction" },
{ "type": "text", "page": 1, "content": "Text content from this section..." },
{ "type": "image", "page": 1, "path": "images/img_001.png" },
{ "type": "table", "page": 3, "path": "images/table_001.png", "context": {
"before": ["Previous text for context..."],
"after": ["Following text..."]
}}
]
}
inventory:
source_files:
- name: presentation.pdf
type: pdf
pages: 12
extracted: true
images:
- path: images/page_01_img_01.png
source: presentation.pdf
page: 1
description: "Architecture diagram"
- path: images/existing/logo.png
source: user_provided
description: "Company logo"
text_blocks: 30
tables: 2
total_images: 7
digraph material_analysis {
"User provides materials" [shape=doublecircle];
"Create raws/ directory" [shape=box];
"Has PDF?" [shape=diamond];
"Decompose PDF" [shape=box];
"Has URL?" [shape=diamond];
"Crawl URL" [shape=box];
"Organize other files" [shape=box];
"Generate inventory.yaml" [shape=box];
"Report to user" [shape=box];
"Proceed to STEP-2" [shape=doublecircle];
"User provides materials" -> "Create raws/ directory";
"Create raws/ directory" -> "Has PDF?";
"Has PDF?" -> "Decompose PDF" [label="yes"];
"Has PDF?" -> "Has URL?" [label="no"];
"Decompose PDF" -> "Has URL?";
"Has URL?" -> "Crawl URL" [label="yes"];
"Has URL?" -> "Organize other files" [label="no"];
"Crawl URL" -> "Organize other files";
"Organize other files" -> "Generate inventory.yaml";
"Generate inventory.yaml" -> "Report to user";
"Report to user" -> "Proceed to STEP-2";
}
When decomposing PDFs, the following extraction patterns apply:
page_{NN}_img_{NN}.pngtable_{NN}.pngIf user provides only images, videos, or other files without a PDF:
Create directory structure:
raws/
images/existing/
videos/
audio/
Copy files to appropriate directories
Create inventory.yaml manually:
inventory:
source_files: []
images:
- path: images/existing/diagram.png
source: user_provided
description: "User-provided diagram"
text_blocks: 0
tables: 0
total_images: 1
| Source | Description | Availability |
|---|---|---|
user_provided | Files directly supplied by the user | Available in raws/images/existing/ |
pdf_extracted | Images and tables extracted from PDF documents | Available after decomposition |
retrieve | Stock images to be fetched from Pexels/Pixabay | Resolved in STEP-4 |
generate | Custom images to be generated via Caro LLM API | Resolved in STEP-4 |
url_crawled | Images downloaded from crawled web pages | Available after crawling |
材料分析完成。
来源: presentation.pdf (12 页)
提取结果:
文本块: 30
图片: 5
表格: 2
输出目录: raws/
- data.json (结构化内容)
- images/ (7 个文件)
- inventory.yaml (材料清单)
请确认材料是否完整，然后继续下一步。
材料整理完成。
用户提供:
- 3 张图片 -> raws/images/existing/
- 1 个视频 -> raws/videos/
- presentation.pdf (12 页)
PDF 提取结果:
文本块: 30
图片: 5
总计: 9 个可用素材
材料清单: raws/inventory.yaml
请确认后继续。
| Issue | Cause | Solution |
|---|---|---|
| PDF images blurry | Low resolution in source PDF | Use higher quality source or request original images |
| Tables not detected | Complex table layout | Extract tables manually as screenshots |
| Text encoding errors | Non-UTF8 PDF encoding | Re-export PDF with UTF-8 encoding |
| Missing sections | Title pattern mismatch | Adjust title detection parameters when invoking decompose tool |
page_{NN}_img_{NN}.pngtable_{NN}.pngexisting/ subdirectories增量模式用于在已有素材库基础上添加新材料，不重新解析已处理的文件。
amendment 字段指定的新素材文件raws/inventory.yaml，了解已有素材清单inventory.yaml，追加新素材条目，保留已有条目不变raws/inventory.yaml