Style Analyzer

文本风格分析器 - 分析写作风格特征并生成 Voice Profile 配置文件。当需要捕捉作者风格、创建 Voice Profile 或分析文本特征时使用。

Security Scan

The skill mostly does what it says (local style analysis), but it also contains an undocumented LLM client that sends text to an external DashScope API and requires a DASHSCOPE_API_KEY which is not declared in the skill metadata or SKILL.md — this mismatch is concerning.

Purpose & Capability

The included scripts implement a style analyzer and an LLM-backed analyzer, which align with the skill name/description. However, the SKILL metadata and SKILL.md present the tool as a local analyzer and omit mention of the LLM mode and its external API dependency, making the package's full capabilities under-documented.

Instruction Scope

SKILL.md documents CLI use for the local script and references pip install of requirements, but does not mention the analyze_style_llm.py script, the need for an API key, or that text (up to 8000 characters) will be sent to an external service. The presence of an LLM client that transmits user text to a remote endpoint expands the runtime scope beyond what's documented.

Install Mechanism

There is no installer (instruction-only), and SKILL.md tells users to run pip install -r scripts/requirements.txt. requirements.txt includes requests, rich, PyYAML — this is expected, but inclusion of requests enables network calls and should have been documented explicitly.

Credentials

Registry metadata lists no required environment variables, but scripts/analyze_style_llm.py requires DASHSCOPE_API_KEY and will raise an error if it's missing. The skill asks to send user text to https://coding.dashscope.aliyuncs.com/v1/chat/completions using that key. An undeclared credential requirement is a proportionality and transparency problem.

Persistence & Privilege

The skill does not request always:true or other elevated persistence. It is user-invocable and can be run manually; there is no evidence it modifies other skills or agent-wide settings.

What to consider before installing

This package contains two modes: a local analyzer (analyze_style.py) and an LLM-backed analyzer (analyze_style_llm.py) that sends text to an external DashScope API and requires an environment variable DASHSCOPE_API_KEY. The skill metadata and SKILL.md fail to declare that API usage or the required env var. Before installing or running: (1) If you only want local analysis, run scripts/analyze_style.py and avoid installing or running analyze_style_llm.py. (2) If you plan to use the LLM script, treat any text you send as being transmitted to a third party — avoid sending sensitive or proprietary text, and verify the API endpoint and the scope of the API key. (3) Audit or remove the LLM script if you need a purely offline tool. (4) Consider updating SKILL.md or metadata to declare DASHSCOPE_API_KEY and the remote call; verify requests in requirements.txt are intentional. (5) Run the tool in an isolated environment (or sandbox) and review network activity if you are unsure.

Style Analyzer - 文本风格分析器

Overview

分析文本的写作风格特征，自动生成 Voice Profile 配置文件，包含风格标签、语速、语气、情感倾向等。用于保持小说创作的风格一致性。

使用场景：

• 需要捕捉作者的写作风格
• 需要为小说项目创建 Voice Profile
• 需要分析文本的句式/词汇/节奏特征
• 需要量化风格特征用于后续生成

分析维度

维度	分析内容	输出
Voice Profile	风格标签、语速、语气、情感倾向	labels, pace, tone, sentiment
句式特征	平均句长、长短句比例、修辞手法	average_length, ratios, rhetorical_devices
词汇特征	词性比例、高频词、词汇丰富度	adj/verb/noun_ratio, frequent_words
节奏特征	段落长度、标点分布、对话比例	paragraph_length, punctuation, dialogue_ratio

CLI 使用

# 基本用法
python3 scripts/analyze_style.py --input novel.txt --output style.yml

# 指定样本大小
python3 scripts/analyze_style.py \
  --input novel.txt \
  --output style.yml \
  --sample-size 2000

# 分析结果输出到终端 + 保存 YAML
python3 scripts/analyze_style.py -i text.txt -o style.yml

参数说明

参数	必填	说明
--input / -i	✅	输入文本文件路径
--output / -o	✅	输出 YAML 配置文件路径
--sample-size / -s	❌	分析样本大小（默认 1000 字符）

输出 YAML 格式

voice_profile:
  labels:
  - 简洁
  - 快速
  - 紧凑
  - 对话性强
  pace: 快速
  tone: 随意
  sentiment: 0.15

sentence_features:
  average_length: 28.5
  long_sentence_ratio: 0.15
  short_sentence_ratio: 0.65
  rhetorical_devices:
    metaphor: 3
    simile: 2

vocabulary_features:
  adj_ratio: 0.12
  verb_ratio: 0.18
  noun_ratio: 0.35
  most_frequent_words:
  - - 主角
    - 15
  - - 说道
    - 12
  unique_words_ratio: 0.68

rhythm_features:
  average_paragraph_length: 156
  punctuation_distribution:
    period: 45
    comma: 78
    question: 12
    exclamation: 5
  dialogue_ratio: 0.35

风格标签说明

标签	含义	触发条件
简洁	句子简短	平均句长 ≤ 40 字符
复杂	句子较长	平均句长 > 40 字符
快速	节奏紧凑	平均句长 < 30 字符
舒缓	节奏缓慢	平均句长 ≥ 30 字符
紧凑	标点密集	每句平均字符 < 30
娓娓道来	叙述舒缓	每句平均字符 ≥ 30
互动性	有问句	包含？或!
对话性强	对话多	包含引号对话
节奏感强	标点丰富	包含，、；

语速分类

语速	条件
缓慢	平均句长 > 50 字符
适中	平均句长 30-50 字符
快速	平均句长 < 30 字符

语气分类

语气	条件
正式	正式标记词 > 随意标记词 × 2
随意	随意标记词 > 正式标记词 × 2
中性	介于两者之间

正式标记词：因此、然而、综上所述、基于此、由此可见

随意标记词：我觉得、我想、你知道、其实、就是说

情感倾向

范围：-1.0 到 1.0

• 正值：积极情感（好、美、喜欢、爱、快乐等）
• 负值：消极情感（坏、丑、讨厌、恨、悲伤等）
• 0 附近：中性情感

依赖

• Python 3.8+
• rich (终端渲染)
• PyYAML (配置文件解析)

安装依赖：

pip install -r scripts/requirements.txt

与其他技能集成

与 smart-prompt-builder 集成

# 1. 分析风格
python3 ../style-analyzer/scripts/analyze_style.py \
  --input reference_text.txt \
  --output style.yml

# 2. 使用生成的 Voice Profile 构建提示词
python3 ../smart-prompt-builder/scripts/build_prompt.py \
  --scene-type description \
  --style-file style.yml \
  --context '{"scene": "森林"}'

与 novel-writer 集成

# 1. 分析参考文本风格
python3 ../style-analyzer/scripts/analyze_style.py \
  --input author_sample.txt \
  --output style.yml

# 2. novel-writer 使用 style.yml 保持风格一致

注意事项

• 样本大小建议 ≥ 500 字符，以获得准确的风格分析
• 支持 UTF-8 和 GBK 编码的文本文件
• 输出 YAML 可直接用于 smart-prompt-builder 的 --style-file 参数
• 修辞手法检测基于关键词匹配，可能不完全准确