sentence-split-zh

name	sentence-split-zh
description	Chinese sentence segmentation with stable numbering. / 中文句子切分与稳定编号。
metadata	{"builtin_skill_version":"0.1.0","copaw":{"emoji":"🧵","requires":{}}}

sentence-split-zh

Purpose / 目标

Split Chinese text into sentence units that are suitable for alignment, diff, and proofreading review.

将中文文本切分为适合对齐、差异分析和校对的句子单元。

When to use / 何时使用

• "Split this chapter into sentence list."
• "把这段按句子拆开并编号。"
• "Need sentence ids before alignment."

Inputs / 输入

• text: required
• keep_paragraph: optional, true or false
• punctuation_mode: optional, strict or tolerant

Output / 输出

• sentences: ordered list with sentence_id
• notes: merge or split decisions for tricky punctuation
• anomalies: suspiciously long or fragmented segments

Boundaries / 边界

• Preserve original order.
• Keep quotations and parentheses coherent when possible.
• If segmentation confidence is low, flag it explicitly.

Execution Guidance / 执行指引

1. Normalize punctuation variants.
2. Segment using sentence-ending punctuation and context cues.
3. Resolve edge cases such as abbreviations or nested quotes.
4. Return numbered sentences with concise notes.

Prompt Contract / 提示约定

• Prefer stable split points over aggressive splitting.
• Keep output easy to map back to source text.

Example I/O / 示例输入输出

Input example:

• task: "Apply this skill to the provided text block"
• constraints: "Keep original intent and only apply necessary changes"

输入示例：

• 任务："对给定文本应用本技能"
• 约束："保持原意，仅做必要修改"

Output example:

• result: concise, structured outcome
• highlights: key findings and priority actions
• follow_up: unresolved items for human review

输出示例：

• 结果：结构化且简洁的处理结果
• 重点：关键发现与优先处理项
• 后续：需人工确认的未决事项