agentsclimarketplace

Readme init

Skill YuAICode/ai-skills/skills/readme-init

35 个即用 Claude Code skill,每个都带离线测试 + skill-doctor 自校验 · 35 tested, self-linted Claude Code skills (中文优先)

Install
npx -y skills add YuAICode/ai-skills --skill readme-init

Assembled from the repository path, not quoted from the project. Check it against their README if it does not work.

One thing to look at

  • 1 stars1 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.

What its author says it does

Copied from the file, not written here

扫描项目自动生成或刷新 README(检测技术栈/结构/可运行脚本)。用户说「/readme-init」「帮我生成 README」「项目没有 README,帮我写一个」「刷新/更新 README」「README 太旧了帮我刷新」时触发。

SKILL.md

3.2 KB, as published. Nobody here has run it

readme-init — 项目 README 自动生成 / 刷新

扫描目标项目的技术栈、入口文件、可运行脚本、目录结构,据此让 Claude 撰写或刷新一份中文 README。 核心原则:只写扫描到的真实事实,拿不准的标「待补」,不臆造功能或命令

何时触发

用户说以下任意一种:

  • /readme-init/readme-init <项目目录>
  • "帮我生成 README" / "这个项目没有 README,帮我写一个"
  • "刷新/更新 README" / "README 太旧了/不准了"
  • "给这个仓库补一份文档"

用法

1. 扫描项目

bash <skill>/bin/scan-project.sh [项目目录]
# 默认当前目录;输出四个段落:
#   LANGUAGES   — 检测到的编程语言/运行时
#   SCRIPTS     — package.json / Makefile / composer.json 中可运行命令
#   ENTRYPOINTS — 存在的常见入口文件
#   STRUCTURE   — 顶层目录与关键文件(忽略 node_modules/.git/dist/build 等)

2. 流程

  1. 扫描:跑 bin/scan-project.sh <项目目录>,拿到结构化事实。
  2. 判断是否已有 README:
    • 无 README:按下方模板从零生成。
    • 已有 README:先读原文,在尊重原有内容的基础上,仅补缺/更新与扫描结果不符的部分,不全量覆盖。
  3. 写 README:依据扫描事实填充模板各节。拿不准的内容一律标 (待补)
  4. 写入文件:默认写到 <项目目录>/README.md;若已有 README 则在确认后覆盖。
  5. 报告:列出已填写的章节、标为「待补」的项目。

README 模板(新建时使用)

# <项目名>

> <一句话简介>(待补:请补充项目用途)

## 技术栈

- <语言/框架>(来自 LANGUAGES 扫描)

## 安装

```bash
# 待补:根据项目实际安装步骤填写

运行

# 来自 SCRIPTS 扫描,如:
# npm run build
# make test

项目结构

<来自 STRUCTURE 扫描,列顶层目录与关键文件>

许可

(待补:未检测到 LICENSE 文件,请补充)


## 硬规则

1. **基于事实**:只写扫描到的语言、脚本、文件。未扫描到的功能/命令不得出现。
2. **拿不准就标「待补」**:项目简介、作者、许可等无法自动推断的内容,一律用 `(待补)` 占位。
3. **已有 README 尊重原文**:补缺优先,改动最小化;不删除原有人工内容。
4. **不执行脚本**:只读取/解析文件,不运行任何项目构建或测试命令。
5. **中文输出**:生成的 README 使用简体中文;代码块、路径、命令保持英文。

## 边界

- `bin/scan-project.sh` 纯 bash + grep,无需 jq/python/node 等外部工具,离线可用。
- 不自动 push/commit;写完后告知用户文件路径,由用户决定是否提交。
- 不处理多 repo monorepo 的子包递归扫描;只扫单一项目根目录。
- 框架/语言靠清单文件推断,无清单文件的项目会尝试文件扩展名兜底(maxdepth 3)。

Gives 0 of the 12 instructions most project setup skills give

Counted across 999 of the 1,637 authors here whose files we hold, read 2026-08-06

  • ask one question at a timein 29 of 999, across 28 files
  • detect the package manager from lockfilesin 28 of 999, across 9 files
  • present findings to the userin 25 of 999, across 4 files
  • explore current repo statein 24 of 999, across 3 files
  • update the agent skills block in place if it existsin 24 of 999, across 3 files
  • install husky lint-staged and prettierin 23 of 999, across 4 files
  • create the lintstagedrc filein 22 of 999, across 3 files
  • commit all changed filesin 22 of 999, across 3 files
  • run lint-staged to verify it worksin 22 of 999, across 3 files
  • initialize huskyin 21 of 999, across 2 files
  • create the husky pre-commit filein 21 of 999, across 2 files
  • create a prettierrc file if missingin 21 of 999, across 2 files

Said here and by no other author read

  • mark uncertain content as pending
  • output simplified Chinese text
  • fill template using only scanned facts
  • respect original content when updating
  • minimize changes to existing README
  • report filled sections and pending items

Grouped from the skills themselves: near-identical wordings counted once, and counted by distinct author, so one author publishing three of these counts once.

Keep looking

Skills are one crate of 328,083. Ordering is by how many stacks a row turns up in, so the top of any crate is what has actually been picked rather than what has the most stars.