← Back to Skills Marketplace
huangliujiao-tal

editor-classify-and-parse

by huangliujiao-tal · GitHub ↗ · v1.0.0 · MIT-0
cross-platform ✓ Security Clean
32
Downloads
0
Stars
0
Active Installs
1
Versions
Install in OpenClaw
/install editor-classify-and-parse
Description
自动识别输入内容的剧本层级(-1到4),并提取场景、角色、台词及动作等结构信息以支持后续处理。
README (SKILL.md)

Skill: editor-classify-and-parse — 输入分类与解析

用途

解析任意形式的输入内容,识别输入层级(Level -1~4),提取剧本结构,为后续创作流程提供分流决策依据。

触发时机

  • 阶段 2:识别输入层级 + 解析剧本结构

核心能力:Input Level Classification(输入层级分类)

Level 定义与判定标准

Level 名称 判定标准 关键特征
-1 前提不成立 Story Premise 四要素缺失或无张力 Want=Need、Stake 为零、无核心戏剧问题
0 空任务号 缺少本集实质性剧情内容 仅有系列描述/前集剧本等参考背景
1 故事大纲 纯叙述,无场景划分,无台词格式 连续文字段落,叙述性语言为主
2 故事剧本 有场景划分和台词,无分镜 存在场景头标、角色对白,但无镜头描述
3 粗稿分镜 有场景和镜头意图描述,缺少细节 存在镜头/画面意图关键词,但描写粗糙
4 详细稿 已有完整场景/台词/动作描述 场景、台词、动作描写均完整

判定流程

1. 输入存在吗?
   ├─ 否 → Level 0
   └─ 是 ↓

2. Story Premise 四要素是否成立?
   ├─ 否 → Level -1(前提不成立)
   └─ 是 ↓

3. 输入中有场景划分(场景头标)吗?
   ├─ 否 → Level 1(故事大纲)
   └─ 是 ↓

4. 输入中有台词(对白行)吗?
   ├─ 否 → Level 1(仍是大纲,只是分了段)
   └─ 是 ↓

5. 输入中有镜头/画面意图描述吗?
   ├─ 否 → Level 2(故事剧本)
   └─ 是 ↓

6. 场景/台词/动作描写是否完整?
   ├─ 否 → Level 3(粗稿分镜)
   └─ 是 → Level 4(详细稿)

核心能力:Script Structure Extraction(剧本结构提取)

对 Level 2+ 的输入,提取以下结构信息:

提取项

  1. 场景列表:所有场景头标 + 时间/室内外标记
  2. 角色列表:所有出场角色 + 首次出场位置
  3. 台词统计:每个角色的台词行数占比(用于声纹校验密度评估)
  4. 动作描写密度:△ 行数 vs 台词行数比值(检测是否满足"每3句台词内1行△"规则)
  5. 道具/场景元素:所有提到的道具、宠物、特殊物品
  6. 变装/特殊形象:角色变装标注是否规范

输出格式

{
  "detected_level": 2,
  "level_confidence": 0.9,
  "level_evidence": "检测到7个场景头标、23句对白、0处镜头意图描述",
  "structure": {
    "scene_count": 7,
    "character_list": ["张飞", "关羽", "诸葛亮"],
    "first_appearances": {"张飞": "场景1", "关羽": "场景2", "诸葛亮": "场景3"},
    "dialogue_stats": {"张飞": 8, "关羽": 6, "诸葛亮": 9},
    "action_density": 0.4,
    "props": ["羽扇", "长矛", "战马"],
    "costumes": []
  },
  "risk_flags": ["动作描写密度不足(0.4 \x3C 0.5 阈值)"]
}

与其他 Skill 的协作

  • editor-story-architect:classify-and-parse 提供 detected_level,story-architect 提供 premise 校验结果,两者共同决定执行计划
  • editor-consistency-checker:提取的角色列表和道具列表供一致性检查使用
  • editor-dialogue-craft:台词统计用于评估声纹校验的工作量
  • editor-format-script-v4:结构提取结果指导格式转换时的模块映射

禁止事项

  • ❌ 不得在分类阶段修改任何输入内容——只读取、分析、判定
  • ❌ 不得跳过 Level 判定直接进入创作——Level 决定后续所有流程的分流
  • ❌ Level 判定不得模棱两可——必须给出明确的 Level 和 confidence,confidence \x3C 0.7 时上报 help_requested
  • ❌ 不得将 Level 3/4 的输入降级为 Level 1 处理——高级别输入的台词保护规则更严格
Usage Guidance
Safe to install for Chinese-language script classification workflows. Users working in other languages should verify outputs carefully or add language-handling guidance, but the skill itself does not ask the agent to modify files, run commands, store data, or contact external services.
Capability Assessment
Purpose & Capability
The stated purpose is input-level classification and script structure extraction, and the artifact content stays aligned with that purpose.
Instruction Scope
Instructions are limited to reading/analyzing supplied text and producing structured results; the skill is written in Chinese and does not define non-Chinese fallback behavior, which is a usability limitation rather than a security concern.
Install Mechanism
The inspected package contains only a Markdown skill file, with no executable scripts, dependencies, install hooks, or package code.
Credentials
The skill does not request filesystem, shell, network, account, credential, or external-service access.
Persistence & Privilege
No persistence, background execution, privilege escalation, local indexing, or long-running worker behavior is described.
How to Use
  1. Make sure OpenClaw is installed (local or Docker)
  2. Run the install command in chat: /install editor-classify-and-parse
  3. After installation, invoke the skill by name or use /editor-classify-and-parse
  4. Provide required inputs per the skill's parameter spec and get structured output
Version History
v1.0.0
初始版本:输入层级分类(Level -1~4) + 剧本结构提取
Metadata
Slug editor-classify-and-parse
Version 1.0.0
License MIT-0
All-time Installs 0
Active Installs 0
Total Versions 1
Frequently Asked Questions

What is editor-classify-and-parse?

自动识别输入内容的剧本层级(-1到4),并提取场景、角色、台词及动作等结构信息以支持后续处理。 It is an AI Agent Skill for Claude Code / OpenClaw, with 32 downloads so far.

How do I install editor-classify-and-parse?

Run "/install editor-classify-and-parse" in the OpenClaw or Claude Code chat to install it in one step — no extra setup required.

Is editor-classify-and-parse free?

Yes, editor-classify-and-parse is completely free, licensed under MIT-0. You can download, install and use it at no cost.

Which platforms does editor-classify-and-parse support?

editor-classify-and-parse is cross-platform and runs anywhere OpenClaw / Claude Code is available (cross-platform).

Who created editor-classify-and-parse?

It is built and maintained by huangliujiao-tal (@huangliujiao-tal); the current version is v1.0.0.

💬 Comments