AgentSkillsCN

md-section-splitter

使用 skills/md-section-splitter/scripts/md_section_splitter.rb，将归一化的 Markdown 按 H2 标题拆分为独立的章节，并生成各章节对应的文件及索引元数据。当您需要对内容进行确定性的章节划分，以便后续提取或审阅时，此方法堪称理想之选。

SKILL.md
--- frontmatter
name: md-section-splitter
description: Split normalized Markdown into H2 sections using skills/md-section-splitter/scripts/md_section_splitter.rb, producing per-section files and index metadata. Use when you need deterministic sectionization for extraction or review.

MD Section Splitter

Overview

Split normalized Markdown into H2 sections while preserving code fences and output an index for downstream processing.

Quick Start

bash
# list sources and their last normalized path
skills/md-section-splitter/scripts/md_section_splitter.rb list

# split all sources in state.json
skills/md-section-splitter/scripts/md_section_splitter.rb split --all

# split a single source by id
skills/md-section-splitter/scripts/md_section_splitter.rb split --id best-practices

Inputs

  • •data/doc-fetcher/state.json (last normalized path per source)
  • •Normalized Markdown under data/doc-fetcher/normalized/<id>/
  • •Subcommands: list, split
  • •Optional flags for split: --all, --id, --force, --dry-run

Outputs

  • •Sections: data/doc-fetcher/sections/<id>/<snapshot_sha>/
  • •Index: data/doc-fetcher/sections/<id>/<snapshot_sha>/index.json
  • •State updates: data/doc-fetcher/state.json

Workflow

  1. •Ensure normalization has run (md-normalizer).
  2. •Run the splitter with split --all or split --id.
  3. •Confirm section files and index.json exist.

Options

  • •list: Print sources and last normalized path (from state.json).
  • •split --force: Overwrite existing section output.
  • •split --dry-run: Do not write files.

Notes

  • •H2 is treated as a section boundary; code fences are respected.
  • •A wrapper exists at scripts/md_section_splitter.rb.