broll-catalog-builder

Compare original and translation side by side

🇺🇸

Original

English

🇨🇳

Translation

Chinese

B-roll Catalog Builder

B-roll目录构建工具

Follow shared release-shell rules in:

```
postplus-shared
```
release-shell rules

Use this skill when the user has B-roll assets and needs a reusable asset inventory before planning the edit.

This skill should create the asset mother file for the B-roll side of the workflow.

Default output:

```
broll-catalog.json
```

Optional supporting outputs:

contact sheets
frame manifests
markdown review summaries

遵循以下共享的release-shell规则：

```
postplus-shared
```
release-shell规则

当用户拥有B-roll素材且需要在规划剪辑前创建可复用的素材库存时，可使用此skill。

此skill应创建工作流中B-roll侧的素材母文件。

默认输出：

```
broll-catalog.json
```

可选辅助输出：

联系表
帧清单
Markdown审核摘要

Use For

适用场景

organizing a B-roll folder into a reusable catalog
tagging B-roll by semantic meaning instead of filename only
identifying the best usable ranges within longer B-roll clips
separating strong proof clips from weak filler clips
preparing assets for later beat-to-B-roll matching

将B-roll文件夹整理为可复用目录
按语义而非仅按文件名给B-roll打标签
识别长B-roll片段中最佳可用范围
将有力佐证素材与弱填充素材区分开
为后续节拍与B-roll匹配准备素材

Trigger Signals

触发信号

Use this skill when the user asks for things like:

先整理一下这些 B-roll 素材
给这些素材打标签
哪些素材适合做 proof
给我一个可复用的 B-roll 素材表
先把素材库建起来

Do not use this skill when the user already has a good catalog and needs final beat matching.

Route that to:

skills/40-creative/b-roll-family/broll-match-engine

当用户提出以下类似需求时使用此skill：

先整理一下这些 B-roll 素材
给这些素材打标签
哪些素材适合做 proof
给我一个可复用的 B-roll 素材表
先把素材库建起来

当用户已有完善的目录并需要最终节拍匹配时，请勿使用此skill。

此时应转向：

skills/40-creative/b-roll-family/broll-match-engine

Read These References

参考资料

family architecture:
```
../references/skill-architecture.md
```
shared contracts:
```
../references/contracts.md
```

Use these existing local skills when helpful:

```
skills/40-creative/frame-extraction
```
a dedicated visual analysis workflow

家族架构：
```
../references/skill-architecture.md
```
共享契约：
```
../references/contracts.md
```

如有需要，可使用以下现有本地skill：

```
skills/40-creative/frame-extraction
```
专用视觉分析工作流

Core Principle

核心原则

Do not catalog assets only by what is literally visible.

Also capture:

what the asset proves
what edit role it can play
how long it stays fresh on screen
whether it is strong enough for primary proof or only support

The same clip may be:

strong proof
weak proof but good transition cover
pure pace reset
UI demo

depending on how it is used.

请勿仅按素材表面可见内容进行编目。

还需记录：

素材能佐证的内容
它在剪辑中可扮演的角色
它在屏幕上能保持新鲜感的时长
它是否足够作为主要佐证素材，还是仅能作为辅助素材

同一素材可能：

是有力佐证素材
是弱佐证但适合作为过渡衔接素材
纯粹用于调整节奏
用于UI演示

具体取决于使用方式。

Required Inputs

必填输入

At least one of:

local folder of B-roll files
manifest of B-roll files
list of local file paths

Optional but valuable:

campaign context
transcript or script
target platform
known proof goals

至少满足以下一项：

B-roll文件的本地文件夹
B-roll文件清单
本地文件路径列表

可选但有价值的输入：

营销活动背景
转录稿或脚本
目标平台
已知的佐证目标

Workflow

工作流

1. Recover the asset set

1. 收集素材集

Build the working file list from:

folder scan
manifest
explicit file list

Preserve local paths.

从以下来源构建工作文件列表：

文件夹扫描
清单
明确的文件列表

保留本地路径。

2. Inspect the assets

2. 检查素材

For each asset, capture:

media type
duration if video
literal description
strongest usable ranges
semantic tags
support roles
visual risks

Use lightweight extraction or contact-sheet generation where helpful.

针对每个素材，记录：

媒体类型
视频时长（若为视频）
字面描述
最佳可用片段范围
语义标签
辅助角色
视觉风险

必要时可使用轻量提取或生成联系表。

3. Judge edit usefulness

3. 判断剪辑实用性

For each asset, decide:

is it primary proof, supporting proof, or filler
what claims it can support
whether it fits 9:16, 16:9, or both
whether text or UI is readable enough

针对每个素材，确定：

它是主要佐证素材、辅助佐证素材还是填充素材
它能支持哪些主张
是否适配9:16、16:9或两者都适配
文本或UI是否足够清晰可读

4. Package the catalog

4. 打包目录

Build

broll-catalog.json

using the shared contract.

The catalog should be reusable across multiple edit runs.

使用共享契约构建

broll-catalog.json

。

该目录应可在多次剪辑流程中复用。

Minimal Script

最小脚本

First implementation script:

```
scripts/run_build_broll_catalog.mjs
```

Current first-version behavior:

scans a local folder recursively
keeps video and image assets
uses
```
ffprobe
```
for duration and dimensions
reads same-basename
```
.md
```
sidecars when present
infers conservative semantic tags and support roles from filename and sidecar text
writes
```
broll-catalog.json
```

Example:

bash

node ${CLAUDE_SKILL_DIR}/scripts/run_build_broll_catalog.mjs \
  --input-dir /abs/path/to/b-roll-source \
  --output <work-folder>/broll-catalog.json

首个实现脚本：

```
scripts/run_build_broll_catalog.mjs
```

当前第一版本的功能：

递归扫描本地文件夹
保留视频和图片素材
使用
```
ffprobe
```
获取时长和尺寸
若存在同名
```
.md
```
附属文件则读取其内容
根据文件名和附属文件文本推断保守的语义标签和辅助角色
生成
```
broll-catalog.json
```

示例：

bash

node ${CLAUDE_SKILL_DIR}/scripts/run_build_broll_catalog.mjs \
  --input-dir /abs/path/to/b-roll-source \
  --output <work-folder>/broll-catalog.json

Release-Shell Execution Contract

Release-Shell执行契约

keep folder scans, contact-sheet inputs, frame manifests, and intermediate catalog builds under
```
<work-folder>/.postplus/broll-catalog-builder/
```
keep only final user-facing
```
broll-catalog.json
```
or review exports outside
```
.postplus/
```
start with a bounded first pass on one folder or shortlist before broader cataloging
if
```
ffprobe
```
or
```
ffmpeg
```
is missing, the user's agent should proactively install the
```
ffmpeg
```
package with the host package manager already present on the machine before continuing
rerun direct checks such as
```
ffprobe -version
```
and
```
ffmpeg -version
```
after installation
if installation or verification fails, stop immediately instead of switching to ad hoc shell glue

将文件夹扫描、联系表输入、帧清单和中间目录构建文件存储在
```
<work-folder>/.postplus/broll-catalog-builder/
```
下
仅将最终面向用户的
```
broll-catalog.json
```
或审核导出文件放在
```
.postplus/
```
外部
在进行更广泛的编目前，先对单个文件夹或候选列表进行有限的首轮处理
若缺少
```
ffprobe
```
或
```
ffmpeg
```
，用户的Agent应主动使用机器上已有的主机包管理器安装
```
ffmpeg
```
包，然后再继续
安装后重新运行直接检查，如
```
ffprobe -version
```
和
```
ffmpeg -version
```
若安装或验证失败，立即停止，不要切换到临时shell脚本

Output Standard

输出标准

The catalog is good if a downstream matcher can answer:

what assets exist
which claims they support
which ones are actually strong
where the useful cut ranges are

If the output is only a file list with vague tags, it is not enough.

若下游匹配工具能回答以下问题，则目录合格：

有哪些素材
它们能支持哪些主张
哪些素材真正有效
有用的剪辑范围在哪里

如果输出仅为带有模糊标签的文件列表，则不合格。

First-Version Boundary

第一版本边界

Prefer:

local file scanning
ffmpeg metadata
frame extraction when needed
markdown and JSON packaging

Do not require:

heavy CV pipelines
perfect object recognition
fully automatic clip scoring without review

优先支持：

本地文件扫描
ffmpeg元数据
必要时的帧提取
Markdown和JSON打包

无需支持：

重型CV流水线
完美的物体识别
无需审核的全自动片段评分

Review Rule

审核规则

When confidence is low, say so.

Do not pretend a weak or generic clip is strong proof footage.

It is better to mark an asset as:

```
generic filler
```

than to overstate its usefulness.

当信心不足时，如实说明。

不要将薄弱或通用的素材伪装成有力的佐证素材。

将素材标记为：

```
generic filler
```

比夸大其用途更好。