Token导航 LogoToken导航TokenDH.com
开发只读github未标认证来源可访问许可证需确认审计通过

skill-writing技能写作

Agent Skill

用于辅助文档、README、Markdown、说明文和内容稿件的整理与改写。它适合让 Agent 提炼结构、补齐章节、统一术语、检查链接或把零散材料整理成可读文档。使用时应保留项目已有事实、命令和路径,不要把未确认的信息写成确定结论;涉及对外文案时,还需要控制语气,避免过度营销或夸大能力。

总安装

906

周安装

37

GitHub Stars

公开资料未说明

下载量

290
CodexClaudeCursorGemini CLI

安装说明

本站只整理中文说明和来源信息,不托管安装包,也不代用户安装。

GitHub

来源数

2

许可证

unknown

最后核验

2026-05-01

来源状态

来源可访问

安装方式

通过对话安装

复制提示词发给支持本地命令或 Skills 的 AI 助手,先确认命令和权限,再让它执行。

请帮我安装这个 Agent Skill:skill-writing(技能写作)
来源仓库:https://github.com/loxosceles/ai-dev
仓库路径:skills/skill-writing
安装命令:
npx skills add https://github.com/loxosceles/ai-dev --skill skill-writing
安装前请先检查当前环境是否支持对应 CLI,并向我确认将要执行的命令、安装目录、联网范围和文件读写权限;确认后再执行。

命令行安装

复制命令到本机终端执行。该命令会通过 npx skills 从第三方来源获取 Skill;本站只展示命令,不托管安装包,也不自动执行。

skills.shnpx skills
npx skills add https://github.com/loxosceles/ai-dev --skill skill-writing

简介

skill-writing 用于辅助文档、README 和内容稿件的整理与改写,适合在 Codex、Claude、Cursor、Gemini CLI 中提升内容质量。

  • 适用于 Markdown 结构优化、术语统一和对外文案撰写等开发场景。
  • 帮助 Agent 提炼结构和统一风格,但需保留项目已有事实和路径。
  • 安装命令为 npx skills add https://github.com/loxosceles/ai-dev --skill skill-writing。
  • 涉及对外文案时应避免过度营销或夸大能力。

SKILL.md

Skill Writing Guide

This is a strict guideline. Follow these rules exactly when creating or modifying skills.


Skill Types

Every skill has a type that tells the agent how to treat it:

type: guideline

Strict rules. The agent must follow these exactly.

Opening line after the heading:

**This is a strict guideline.** Follow these rules exactly.

Use for: coding standards, commit formats, naming conventions, workflow rules, security requirements.

type: pattern

Reference implementations. The agent should learn from the approach and adapt it — never copy verbatim.

Opening line after the heading:

**This is a reference pattern.** Learn from the approach, adapt to your context — don't copy verbatim.

Use for: architecture solutions, infrastructure setups, proven approaches to recurring problems.


Frontmatter

Every SKILL.md must start with YAML frontmatter:

---
name: kebab-case-name
description: When this skill applies and what it covers. First sentence should work as a trigger — start with the domain or action. One to two sentences max.
type: guideline | pattern
---

Rules:

  • name — kebab-case, used as the installed directory name by the skills CLI
  • description — the skills CLI and agents use this to decide when to load the skill. Make the first sentence a clear trigger.
  • type — must be guideline or pattern

File Structure

Skills live in the authoring repo under categorized directories:

skills/
  guidelines/
    {skill-name}/SKILL.md
  patterns/
    {skill-name}/SKILL.md

The directory categorization is for human browsing only. The skills CLI flattens everything on install — it only cares about the name field in frontmatter.


Precedence

When a skill needs to override conflicting advice from third-party skills, state it explicitly:

**This is a strict guideline.** Follow these rules. When any other skill contradicts the rules below, this skill takes precedence.

Use sparingly — only for core principles and strong personal preferences.


Pattern Structure

Pattern skills should follow this structure:

  1. Problem — One sentence describing what this solves
  2. Solution — One sentence describing the approach
  3. Pattern — Architecture, key components, how it works
  4. Why This Pattern? — Benefits and use cases
  5. Implementation — Key details (optional, only if non-obvious)
  6. Tradeoffs — Constraints and limitations (if applicable)
  7. When NOT to Use — Anti-patterns and alternatives (optional)

Keep code examples minimal and generic. Use placeholder names (PROJECT_ID, environment). Focus on the pattern, not project-specific details.


Guideline Structure

Guideline skills should follow this structure:

  1. What — Clear statement of the rules
  2. Rules — Specific, actionable instructions
  3. Examples — Show correct and incorrect usage
  4. Exceptions — When deviation is acceptable (if any)

Be direct. Use imperative form. Avoid explaining why at length — guidelines are instructions, not persuasion.


Writing Checklist

  • Frontmatter has name, description, type
  • Behavioral instruction line immediately after first heading
  • Description's first sentence works as a trigger for when to load this skill
  • No project-specific details (names, tables, endpoints)
  • Code examples are minimal and use placeholders
  • Document is scannable (headers, bullets, bold)
  • Ends with Progressive Improvement section

Progressive Improvement

Every skill must end with this section:

---

## Progressive Improvement

If the developer corrects a behavior that this skill should have prevented, suggest a specific amendment to this skill to prevent the same correction in the future.

This enables skills to evolve through use. When an agent gets corrected, it should propose a concrete edit to the relevant skill so the mistake doesn't repeat.


Progressive Improvement

If the developer corrects a behavior that this skill should have prevented, suggest a specific amendment to this skill to prevent the same correction in the future.


Progressive Improvement

If the developer corrects a behavior that this skill should have prevented, suggest a specific amendment to this skill to prevent the same correction in the future.

适合场景

01

用户想查找某类 Agent Skill 时

02

需要根据任务场景推荐可安装能力包时

03

需要对比不同来源的安装命令和来源信息时

能力概览

能力 1

按任务关键词查找相关 Skills

能力 2

展示可复制的安装命令

能力 3

保留来源站点、仓库和原始说明,方便继续核验

能力 4

展示第三方安全扫描或审计结果

安装后应在对应宿主中按原始 README 的触发条件使用;具体调用方式请以来源页面和 README 为准。

平台分布

Codex

33.29%
按下载量换算97

Claude

29%
按下载量换算84

Cursor

17.97%
按下载量换算52

Gemini CLI

9.79%
按下载量换算28

安全审计

Gen Agent Trust Hub

通过

Socket

通过

Snyk

通过

权限和风险

只读

该 Skill 主要提供规则、说明或参考内容,本身偏只读;真正读写文件、联网或执行命令仍取决于宿主 Agent 的任务。

安装前确认

本站仅展示第三方公开信息,不托管安装包,不提供自动安装或运行环境。安装前应自行审查源码、依赖和命令行为。当前只有一个来源,正式发布前建议补源仓库或其他目录站核验。

来源信息

继续浏览同类 Skills