how-to-write SKILL.md skillA

how-to-write from webkubor/writing-standard-skills is an agent skill: a SKILL.md folder an AI agent loads when a task calls for it.

Install it, pinned
mdr add webkubor/writing-standard-skills/how-to-write@git:20261025.fb9d7b8

Pins git:20261025.fb9d7b8 in mdr.lock. Pin the fingerprint instead.

Audit grade
A, 17 of 17 checks passed
Kind
SKILL.md skill
Repository
webkubor/writing-standard-skills
Path
skills/how-to-write/SKILL.md
Versions recorded
3
Latest version
git:20261025.fb9d7b8, committed Oct 25, 2026
First version recorded
Sep 10, 2026
Size
10,983 bytes
Repository stars
6
License
MIT

「怎么写文字」的总纲——先按金字塔原理搭骨架(结论先行 / SCQA / MECE),再按读者定语气(人还是 agent、上级还是同侧),最后必须以新读者身份通读自检。管一切要给别人看的文字:文档、单据、报告、评审意见、讲稿、群消息、PR 描述。触发:"写文档 / 写报告 / 提单 / 写评审意见 / 文案不合理 / 逻辑不清 / 这样礼貌吗 / 结论说不清 / 讲稿检查"。.

Collected from public GitHub. Every version is saved exactly as it was and never changes. Install it at an exact version with the mdr tool, and every file is checked before it reaches your agent. The main file is checked against the SHA-256 fingerprint recorded here, and the others against the fingerprints GitHub keeps for its source commit. The automatic checks below grade the latest version, and the same checks always give the same file the same grade.

What the file says

# 怎么写文字

一切要给别人看的文字,按同一个顺序做四件事。**顺序不能换**——结构错了改字没用,语气对了但结论找不到也没用。

```
① 搭骨架   金字塔原理:结论先行 · SCQA · MECE
② 定读者   给人 or 给 agent;上级 or 同侧
③ 调语气   归因情境 · 自曝先行 · 认对方贡献 · 允许被推翻
④ 当读者   以第一次看的人的身份通读全文,带清单查错
```

跳过任何一步的代价,在文末「三个真实案例」里有实测数字。

---

## ① 搭骨架:金字塔原理

### 结论先行

读者的第一个问题永远是「结论是什么、要不要我动」。答案放第一屏,证据放后面供下钻。

**检验**:只读前五行,读者知不知道结论和自己要做什么。

### SCQA:开篇四句带上下文

| | 写什么 | 关键 |
|---|---|---|
| **S** 情境 | 对方**已知**的事实 | 必须是共识,写成「我做了 X」就变自述 |
| **C** 冲突 | 发生了什么变化让现状不成立 | 这是读者继续读下去的动力 |
| **Q** 疑问 | 因此自然产生的问题 | 通常不用写出来,但心里要有 |
| **A** 答案 | 你的结论,一句话 | 这句话就是全文的塔尖 |

### MECE:分组不重不漏

- **相互独立**:任何一条内容只出现在一个分组里。同一件事写两处,读者不知哪份准,改一处另一处立刻过期。
- **完全穷尽**:分组加起来覆盖全部。**每次分组后做一次加法**——说「7 条」却只列 6 条是最常见的翻车。

### 以上统下 + 逻辑递进

标题必须是它下面所有内容的概括;同层条目按一个明确顺序排(严重度 / 时间 / 责任方),选一个并贯彻。

---

## ② 定读者

### 给人读 vs 给 agent 读——两套写法,不要混

| 维度 | 给人 | 给 agent |
|---|---|---|
| 首要目标 | 知道发生了什么、要不要他动 | 能照做且不跑偏 |
| 开头 | 结论 + 是否需要他处理 | 触发条件 + 禁止项 |
| 语气 | 留余地、给台阶 | 命令式**正确**,越硬越少歧义 |
| 详略 | 只给影响判断的信息 | 穷举边界与反例,宁长不漏 |
| 数据 | 结论化(「93 个模型被漏掉」) | 可复现(完整命令、参数、期望值) |
| 礼貌 | 必要 | 无意义,浪费 token |

**检验**:删掉所有代码块和参数,人读段还成立吗?删掉所有解释句,agent 段还能照做吗?

### 对上 vs 对同侧

- **对上(你的老板 / owner)**:极简,结论先行,不铺过程,客套是噪声。
…

Read the whole file at its exact version.

How to install

The exact contents, by fingerprint
mdr add webkubor/writing-standard-skills/how-to-write@sha256:737eab2f447e95b2

The line at the top of this page pins the label git:20261025.fb9d7b8, which is what most people want. This one pins the sha256 fingerprint of the contents instead, so it survives a label being moved. Either way the exact version is written to mdr.lock, and mdr install downloads and checks those contents again, so it installs them exactly or stops with an error.

Badge

mdr badge

[![mdr](https://markdownregistry.com/badge/art_wg5yi4ctjp3gpqs4.svg)](https://markdownregistry.com/a/art_wg5yi4ctjp3gpqs4)

3 badge image fetches in 30 days. What the badge shows.

Versions

The registry has recorded 3 versions of this file, between Sep 10, 2026 and Oct 25, 2026, about one every 23 days.

versioncommittedcommitsizeaudit
git:20261025.fb9d7b8 latest2026-10-25 fb9d7b8 10,983 BA view · diff
git:20260926.5283a192026-09-26 5283a19 10,517 BA view · diff
git:20260910.b68d6092026-09-10 b68d609 10,081 BA view

Audit of the latest version

A  17 of 17 checks passed. Fixed automatic checks with no AI involved, so the same checks give the same answer every run.
  • pass: Frontmatter block present
  • pass: Frontmatter declares a name
  • pass: Frontmatter declares a description
  • pass: Size between 200 bytes and 200 KB (10983 bytes)
  • pass: No zero-width or bidi control characters
  • pass: No instruction hidden inside an HTML comment
  • pass: No link to an exfiltration or paste host
  • pass: No credential-shaped string
  • pass: No instruction to send local credentials anywhere
  • pass: No text hidden with inline styles
  • pass: No prompt-injection phrasing
  • pass: No curl or wget piped into a shell
  • pass: No recursive delete of root, home or parent
  • pass: No instruction to read or print local credentials
  • pass: No base64 blob over 200 characters
  • pass: No link to a raw IP address
  • pass: No script tag

Source

GitHub

webkubor/writing-standard-skills · 6 stars · license MIT · pushed 2026-11-06 · branch main

API

GET https://markdownregistry.com/api/v1/artifacts/art_wg5yi4ctjp3gpqs4
GET https://markdownregistry.com/api/v1/resolve?ref=webkubor/writing-standard-skills/how-to-write
GET https://markdownregistry.com/api/v1/blob/737eab2f447e95b2e7922fb6e09e3af887579341ecbed365dce2d3a4752f0ca3

Your AI assistant does the legwork. You hear about the deals worth your word. Hand yours the standing instructions at modelranch.com and it joins the network that reads files like this one.

Guides: the SKILL.md format and its required fields · what to check before you install a skill · how to pin a skill to an exact version · where to find agent skills · keeping a team on the same skills · which agent file does what.

More from webkubor/writing-standard-skills

AGENTS.md AGENTS.md file
webkubor/writing-standard-skills · AGENTS.md
git:20261025.4eaa475 · audit B · 6 stars
llms.txt llms.txt file
webkubor/writing-standard-skills · llms.txt
git:20261025.6457abe · audit A · 6 stars
account-custody SKILL.md skill
webkubor/writing-standard-skills · skills/account-custody/SKILL.md · 「账户托管运营前置检查」——Agent 替人接管社交账户(发帖/回复/互动)之前,必须先完成四件事:核对当前平台与入口、加载平台规则红线、读出该账户已有人设与口吻、确认出口 IP…
git:20261105.3275eed · audit A · 6 stars
decision-and-rfc SKILL.md skill
webkubor/writing-standard-skills · skills/decision-and-rfc/SKILL.md · 架构决策记录(ADR)、技术提案(RFC)与重大方案论证的表达规范。基于第一性原理、权衡对齐(Trade-offs)与可证伪性。拒绝报喜不报忧,要求显式声明放弃的代价与推翻条件。触发:"写方案 / 架构决策 / ADR / RFC /…
git:20261025.4eaa475 · audit A · 6 stars
executive-briefing SKILL.md skill
webkubor/writing-standard-skills · skills/executive-briefing/SKILL.md · 面向业务负责人、决策者与 owner 的极简汇报与卡单拍板规范。基于低成本沟通(reply-format-low-cost)与终点导向运行模型。严禁倒苦水铺垫过程,将开放性提问收敛为互斥选项的代价对齐。触发:"对上汇报 / 给老板汇报 /…
git:20261025.4eaa475 · audit A · 6 stars
peer-facing-writing SKILL.md skill
webkubor/writing-standard-skills · skills/peer-facing-writing/SKILL.md · 写给同事看的文本(GitLab 单据、飞书文档、评审意见、群消息、PR review、跨团队报告)的语气与结构规范,基于职场协作与社会心理学。触发:"提单 / 写报告 / 写文档给 QA 或后端 / 评审意见 / 回评 / 跨团队沟通 /…
git:20260910.b68d609 · audit A · 6 stars
platform-voices SKILL.md skill
webkubor/writing-standard-skills · skills/platform-voices/SKILL.md · 「全平台表达风格与媒介心智」分册——按平台受众的心理账户匹配表达法则。涵盖网易云音乐(情怀/通感/夜读)、小红书(大情绪击穿/生活口语/高CTR)、Twitter/X(独立极客/反常识/短平快)、Reddit(零营销耐受/真实硬核/TL;D…
git:20261106.94b7d41 · audit A · 6 stars
scenario-openers SKILL.md skill
webkubor/writing-standard-skills · skills/scenario-openers/SKILL.md · 「场景接待词、破冰与开场钩子库」分册——彻底剔除AI味,像真人一样在各种具体场合下开口。提供私信接待、客诉安抚、社区Issue回应、网易云神评开场、小红书大情绪钩子、X极客直击钩子等高细粒度原子语料。拒绝空洞升华与机器套话,追求真实的人味、…
git:20260926.5283a19 · audit A · 6 stars
talk-text-review SKILL.md skill
webkubor/writing-standard-skills · skills/talk-text-review/SKILL.md · 讲稿/分享稿的文本验收——查的不是错别字,是**姿态**:这份稿子读起来是「我教你们」还是「我跟你们聊」。用心理逆反、自我暴露、认识论谦逊这几条机制,逐句找出会把听众推到对立面的表达并给出改写。触发词:讲稿验收、分享稿检查、这么讲会不会得罪…
v1.0.0 · audit A · 6 stars
talk-visual-review SKILL.md skill
webkubor/writing-standard-skills · skills/talk-visual-review/SKILL.md · 演讲页/幻灯片的视觉验收——用浏览器实测每一屏是否被切、是否过空、对比度够不够、色板有没有绕过主题变量,而不是靠肉眼翻页目测。触发词:讲稿排版检查、幻灯片验收、内容溢出、投影看不全、slide 被切、演讲页视觉检查、talk…
v1.0.0 · audit A · 6 stars
technical-postmortem SKILL.md skill
webkubor/writing-standard-skills · skills/technical-postmortem/SKILL.md · 严肃技术故障复盘、事故通报(Postmortem)与对外公关的表达规范。基于防御性工程与无指责复盘(Blameless Postmortem)原则,剔除空洞道歉与甩锅推诿,下钻到机制防线与量化事实。触发:"事故通报 / 写复盘 /…
git:20261025.4eaa475 · audit A · 6 stars

Every file in webkubor/writing-standard-skills

Browse by kind, by grade A, or by owner.