agentleFS
phanghonghaophanghonghao/THU-Awesome-Skills/skills/repo-docs-zh/SKILL.md

repo-docs-zh

Skill8 starsChanged 3 months agoMITView on GitHub

Generate and maintain repo-docs with Chinese as the primary reader language while preserving source identifiers for lookup. Use when the user asks for Chinese repo docs, mentions repo-docs-zh, wants repo documentation in Chinese, or wants an existing repo-docs package localized for Chinese readers.

The file

We split this skill into the rules it gives an agent

11 rules. About 0 are common, 7 are distinctive, and 4 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈7

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 4

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Use Chinese for titles, explanations, reading paths, examples, caveats, and change-log entries.

    line 27
  2. Distinctive:

    Keep source identifiers exact.

    line 28
  3. Distinctive:

    Use a Chinese reader handle before a source locator.

    line 36
  4. Distinctive:

    Prefer a Chinese visible label that links to the source file over showing the raw path as visible text.

    line 47
  5. Distinctive:

    Use inline code only for short identifiers that matter to the mechanism.

    line 48
  6. Very specific:

    Create references/source-map.md only when many related source locations repeat across pages or need one lookup table.

    line 49
  7. Distinctive:

    Never make a long path the visible link text in narrative prose.

    line 50
  8. Very specific:

    Put dense fields, command lists, schemas, artifacts, and metrics in references/.

    line 72
  9. Very specific:

    Use the main repo-docs English Repo docs block for AGENTS.md and other agent instruction Markdown.

    line 85
  10. Distinctive:

    After first build, reply in Chinese with:

    line 89
  11. Very specific:

    Keep the reply short. Durable guide history belongs in repo-docs/change-log.md.

    line 96

≈ “Common” is an estimate: at least 1 in 5 of the 350,495 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Repo-Docs ZH

Position

This is the Chinese-language overlay for repo-docs. Structure, page ownership, evidence rules, sync behavior, and validation come from ../repo-docs/SKILL.md. This overlay only changes language and Chinese reader experience.

Before acting, read:

  1. ../repo-docs/SKILL.md
  2. ../repo-docs/REFERENCE.md as the topic router only when detailed rules are needed
  3. The routed topic file only when needed
  4. ../repo-docs/EXAMPLES.md only for output shape or tone examples needed by the task

Core Principle

Chinese repo-docs are not English docs translated line by line. They should rebuild the reader's conceptual handles in Chinese.

Chinese carries understanding: what this thing is, why it exists, what happens, and how to check it. English identifiers carry lookup: paths, commands, fields, API names, class/function names, metric names, package names, and dataset names.

Language Rules

  • Use Chinese for titles, explanations, reading paths, examples, caveats, and change-log entries.
  • Keep source identifiers exact. Do not translate paths, commands, keys, API names, class/function names, metric names, error strings, package names, or dataset names.
  • First introduce repeated project terms as 中文名(English term) or 中文名(source identifier); later prefer the Chinese name.
  • README route sections use ## 阅读路径, not ## Reader Routes.
  • Glossary columns are exactly 术语 | 项目里的意思 | 延伸阅读.
  • Narrative page evidence note: 证据状态:除特别标注外,本页基于当前源码已确认。

Chinese Handles And Source Locators

Use a Chinese reader handle before a source locator.

TypeUse in Chinese docs
读者句柄The narrative subject, such as “导入流程”, “会话层”, “结果汇总”, “运行脚本”.
源码定位符Paths, functions, classes, fields, commands, artifact paths. Link directly with a Chinese label when one locator supports one claim.
精确查表名Metrics, schema keys, tool parameters, artifact file names. Usually belongs in references/.
外部术语Terms like benchmark, agent, workspace, protocol, memory. First mention gets a Chinese handle; later prefer Chinese when the term appears in the inspected repo.

Default locator rule:

  • Prefer a Chinese visible label that links to the source file over showing the raw path as visible text.
  • Use inline code only for short identifiers that matter to the mechanism.
  • Create references/source-map.md only when many related source locations repeat across pages or need one lookup table.
  • Never make a long path the visible link text in narrative prose.

Good:

  • 运行脚本负责选择运行模式和输入来源。
  • 会话层先取上下文,再记录本轮真实产生的输出。
  • 汇总器从最终状态读取完成情况和错误原因。

Bad:

  • packages/app/scripts/run_example.sh 是入口。
  • Session.run_task(...) handles context and scheduled observations.
  • ResultCollector.collect(...) records output status.

Readability Rules

  • A paragraph should still make sense after removing backticked identifiers. If not, rewrite around Chinese actions first.
  • Lower code-name density by changing explanation order, not by hiding evidence.
  • Give the reader a concrete handle first: a user action, visible state change, plain concept, or question they already have.
  • 中文技术文档的真人感是具体、直接、敢标未确认;不要为了“有灵魂”加入第一人称、金句或情绪化评价。
  • 成稿前做一次去 AI 表达:删掉“此外”“值得注意”“至关重要”“关键作用”“彰显”等空转词,少用破折号和粗体小标题。
  • 避免让“不仅……而且……”或“不是……而是……”承担主解释;改成正面机制句。
  • Put dense fields, command lists, schemas, artifacts, and metrics in references/.
  • Flowcharts support prose; they do not replace it.
  • A walkthrough step links to the module where a durable concept first matters.

Page Shape Notes

  • README: Chinese opening prose plus links to walkthroughs, references, and glossary. Add ## 阅读路径 only when multiple goals need a table.
  • Walkthrough: numbered ## Step N: 行为名 headings; prose explains mechanism; verification appears once near the end.
  • Module: keep ## 白话模型, ## 代码模型, ## 接下去阅读.
  • Reference: lookup table only; if it starts explaining why behavior exists, move that explanation to a walkthrough or module.

Agent Instruction Block

Use the main repo-docs English Repo docs block for AGENTS.md and other agent instruction Markdown. Chinese mode changes the reader-facing guide, not the agent instruction language.

Build Delivery

After first build, reply in Chinese with:

  1. what readers can now understand or follow;
  2. where to start: repo-docs/README.md -> main walkthrough;
  3. key pages and their jobs;
  4. scope, uncovered areas, validator result, and root agent-file status.

Keep the reply short. Durable guide history belongs in repo-docs/change-log.md.

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.