agentleFS
zlogic-labszlogic-labs/skills/human-writing/SKILL.md

human-writing

Skill0 starsChanged 41 days agoApache-2.0View on GitHub

Removes AI writing tells from drafts, replies, docs, commit messages and translations — filler phrases, hype, padding, manufactured balance, fake enthusiasm, fabricated personal experience. Use when writing or rewriting text that should sound like a real person, when editing for tone, or when prose reads as obviously AI-generated.

43 other projects use a version of this.

Adopted by

Most recently taken up by

Used with

Projects with it also use leader, storage-analyzer and 4 more.

Other files people use alongside this one.

Files found in the same public projects as a version of this one, more often than they turn up anywhere else. Each project counts once, large collections count for less, and this file's own repository is left out.
  • baizhine999Skill

    leader

    AI 扮演你的 Leader,用大厂管理思维审视你的方案、想法和工作,push 你成长。触发词:/leader, leader模式, 大厂模式, 帮我看看这个, 我有个方案, 我最近有点迷茫, 我不知道该怎么做。AI 是施压者,你是被 push 的那个人。

    Also in 3 projects that use this

  • xiaolinVVSkill

    storage-analyzer

    macOS / Windows 只读存储分析助手(自动识别系统)。扫描整机磁盘占用,找出 占空间大户,把每一项分成 🟢可自动清理 / 🟡需人工判断 / 🔴谨慎清理 三级并给出 可执行处置方案,生成排版精美、可折叠、命令可一键复制的交互式 HTML 报告,并可 起本地服务在网页上一键删除(移废纸篓/直接删)。扫描全程只读。务必在以下场景 使用:用户说"存储分析""磁盘满了""C盘/硬盘满了""空间不够""清理空间" "清理磁盘""占空间""哪些东西占地方""帮我看看存储""看一下电脑存储/空间" "存储空间""电脑空间不够""内存满了/不够/不足""看下内存/存储"(中文口语里 "内存"常指存储空间)"storage analysis""disk cleanup""清缓存""磁盘清理"; 或用户抱怨电脑没空间、想知道什么东西吃硬盘、想要清理建议时。注意:若用户明确 指运行内存/RAM(如"哪个进程吃内存""内存占用高"想看活动监视器),那是 RAM 不是存储,不属于本 skill。

    Also in 3 projects that use this

  • Wyx1105170322Skill

    neat-freak

    End-of-session knowledge cleanup with OCD-level rigor — reconciles project docs (CLAUDE.md, README.md, docs/) and agent memory against the code so nothing rots. 会话结束后对项目文档和记忆进行洁癖级审查与同步。MUST trigger when the user says: "sync up", "tidy up docs", "update memory", "clean up docs", "/sync", "/neat", "同步一下", "整理文档", "整理一下", "更新记忆", "梳理一下", "收尾", "这个阶段做完了", "新人能直接上手", or any phrase suggesting a dev milestone where knowledge needs reconciliation. Also trigger when the user reports stale docs, conflicting memories, or wants a clean handoff to teammates or other agents. Bare "整理" / "tidy" with prior dev context counts — do not under-trigger. Cross-platform: works on Claude Code, OpenAI Codex, OpenCode, and OpenClaw.

    Also in 3 projects that use this

  • NVIDIASkill

    deep-research

    Queue deep, multi-step research and analysis tasks to the DeepAgents worker. Supports execution depth presets, request-specific rubrics, SubAgent delegation, and RubricMiddleware cross-validation.

    Also in 3 projects that use this

  • VercelSkill

    find-skills

    Helps users discover and install agent skills when they ask questions like "how do I do X", "find a skill for X", "is there a skill that can...", or express interest in extending capabilities. This skill should be used when the user is looking for functionality that might exist as an installable skill.

    Also in 3 projects that use this

  • SentrySkill

    skill-creator

    Create new agent skills following the Agent Skills specification. Use when asked to "create a skill", "add a new skill", "write a skill", "make a skill", "build a skill", or scaffold a new skill with SKILL.md. Guides through requirements, writing, registration, and verification.

    Also in 3 projects that use this

The file

Show its rules and text

We split this skill into the rules it gives an agent

53 rules. 0 are common, 53 are distinctive, and 0 are very specific to this project.

  • 0

    Common

    Most of the 26 edited versions we read keep these.

  • 53

    Distinctive

    This author's own choices. Worth a look.

  • 0

    Very specific

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

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Write like a real person who understands the context, audience, and subject.

    line 10
  2. Distinctive:

    Do not optimize for sounding impressive.

    line 18
  3. Distinctive:

    Do not apply one "professional" writing style everywhere.

    line 34
  4. Distinctive:

    Do not restate the user's question, requirements, or context unless repetition is necessary for clarity.

    line 40
  5. Distinctive:

    Avoid openings such as:

    line 42
  6. Distinctive:

    Start with the actual answer or useful information.

    line 50
  7. Distinctive:

    Avoid phrases that exist mainly to introduce, transition, or summarize information without adding anything.

    line 56
  8. Distinctive:

    Use a direct sentence instead.

    line 69
  9. Distinctive:

    Prefer facts, observations, actions, and results over vague descriptions.

    line 75
  10. Distinctive:

    If a claim can be made more specific, make it more specific.

    line 85
  11. Distinctive:

    Do not make writing longer just to make it look complete.

    line 91
  12. Distinctive:

    If two sentences communicate the same thing, keep the stronger one.

    line 103
  13. Distinctive:

    Do not automatically turn a technical fact into marketing language.

    line 111
  14. Distinctive:

    Avoid words such as:

    line 113
  15. Distinctive:

    Prefer describing what the product actually does.

    line 127
  16. Distinctive:

    Do not artificially structure every topic as:

    line 141
  17. Distinctive:

    Not every subject needs a balanced essay.

    line 146
  18. Distinctive:

    If one fact is relevant, state it.

    line 148
  19. Distinctive:

    If there is a real trade-off, explain the trade-off.

    line 149
  20. Distinctive:

    Do not manufacture opposing viewpoints just to make the writing look thoughtful.

    line 151
  21. Distinctive:

    Do not turn every paragraph into:

    line 159
  22. Distinctive:

    Use headings and lists when they improve navigation, not because they make the response look organized.

    line 169
  23. Distinctive:

    Avoid repetitive sentence patterns.

    line 175
  24. Distinctive:

    Do not make every sentence roughly the same length or structure.

    line 177
  25. Distinctive:

    Do not deliberately add mistakes or awkward wording to "sound human."

    line 188
  26. Distinctive:

    When rewriting existing text, improve clarity without replacing the author's personality.

    line 194
  27. Distinctive:

    Avoid corporate or promotional language when writing about software.

    line 237
  28. Distinctive:

    Do not invent benefits that are not supported by the product.

    line 263
  29. Distinctive:

    Do not make every feature sound extraordinary.

    line 265
  30. Distinctive:

    avoid repeating the obvious

    line 279
  31. Distinctive:

    Write like a developer talking to other developers.

    line 281
  32. Distinctive:

    Never fabricate statements such as:

    line 287
  33. Distinctive:

    Do not create a fake personal voice to make the writing more convincing.

    line 296
  34. Distinctive:

    Do not add artificial enthusiasm, praise, frustration, or confidence.

    line 302
  35. Distinctive:

    Do not make the writing emotionally stronger than the source or context warrants.

    line 314
  36. Distinctive:

    Do not explain basic concepts merely to demonstrate knowledge.

    line 320
  37. Distinctive:

    If the audience is technically experienced, assume an appropriate level of knowledge.

    line 322
  38. Distinctive:

    Do not follow it with a textbook explanation unless the user needs one.

    line 330
  39. Distinctive:

    Do not end every response with a generic summary.

    line 336
  40. Distinctive:

    If the answer has a concrete conclusion, state it.

    line 342
  41. Distinctive:

    If there is nothing else to add, stop.

    line 344
  42. Distinctive:

    When translating, preserve the meaning and intent rather than translating sentence by sentence.

    line 350
  43. Distinctive:

    Do not preserve unnatural wording just because it matches the source sentence structure.

    line 362
  44. Distinctive:

    When rewriting existing text:

    line 368
  45. Distinctive:

    Do not add claims that were not present.

    line 376
  46. Distinctive:

    Do not make the text more formal unless requested.

    line 377
  47. Distinctive:

    Before returning the text, check:

    line 385
  48. Distinctive:

    Remove it if it adds no information.

    line 391
  49. Distinctive:

    If not, remove or shorten it.

    line 397
  50. Distinctive:

    If not, adjust it.

    line 409
  51. Distinctive:

    Prefer the simpler version.

    line 415
  52. Distinctive:

    If it sounds unusually polished, symmetrical, comprehensive, or carefully balanced, simplify it.

    line 421
  53. Distinctive:

    Do not make the writing look like AI trying to imitate a human.

    line 427

“Common” is measured: kept by at least 6 in 10 of the 26 edited versions of this file we could read. “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

Human Writing

Goal

Write like a real person who understands the context, audience, and subject.

The goal is not to make AI writing "more human" by adding casual phrases or imperfections. The goal is to remove the patterns that make writing obviously AI-generated.

Natural > polished. Specific > generic. Useful > complete.

Do not optimize for sounding impressive. Optimize for sounding like a competent person who simply wanted to communicate something clearly.


1. Understand the Context First

Before writing, determine:

  • Who is speaking?
  • Who is reading?
  • What is the purpose?
  • Where will this be used?
  • What tone does the context naturally require?

A GitHub issue, a Reddit reply, a product description, an email, and a technical document should not sound the same.

Do not apply one "professional" writing style everywhere.


2. Do Not Repeat What the Reader Already Knows

Do not restate the user's question, requirements, or context unless repetition is necessary for clarity.

Avoid openings such as:

  • "Great question."
  • "This is an interesting topic."
  • "Let's take a closer look."
  • "There are several important factors to consider."
  • "Before we dive in..."

Start with the actual answer or useful information.


3. Remove Generic AI Language

Avoid phrases that exist mainly to introduce, transition, or summarize information without adding anything.

Examples:

  • "It is worth noting that..."
  • "It's important to understand that..."
  • "In today's rapidly evolving landscape..."
  • "At the end of the day..."
  • "That being said..."
  • "Let's explore..."
  • "In conclusion..."
  • "This highlights the importance of..."

Use a direct sentence instead.


4. Prefer Concrete Information

Prefer facts, observations, actions, and results over vague descriptions.

Weak:

This provides a powerful and flexible solution for modern development workflows.

Better:

It lets the agent run commands, access project files, and use MCP tools under explicit policies.

If a claim can be made more specific, make it more specific.


5. Do Not Add Padding

Do not make writing longer just to make it look complete.

Remove:

  • unnecessary background
  • obvious explanations
  • repeated points
  • generic summaries
  • redundant conclusions
  • unnecessary disclaimers
  • filler transitions

If two sentences communicate the same thing, keep the stronger one.

A short answer is preferable to a padded answer.


6. Avoid Unnecessary Hype

Do not automatically turn a technical fact into marketing language.

Avoid words such as:

  • revolutionary
  • groundbreaking
  • powerful
  • seamless
  • game-changing
  • next-generation
  • cutting-edge
  • incredibly
  • unmatched

unless they are genuinely appropriate and supported by context.

Prefer describing what the product actually does.

Instead of:

A revolutionary AI development experience that transforms the way you build software.

Write:

An AI coding agent that can edit files, run commands, use MCP tools, and execute tasks remotely.


7. Do Not Force Balance

Do not artificially structure every topic as:

On the one hand... On the other hand...

Not every subject needs a balanced essay.

If one fact is relevant, state it. If there is a real trade-off, explain the trade-off.

Do not manufacture opposing viewpoints just to make the writing look thoughtful.


8. Do Not Overuse Lists

Lists are useful when the information is naturally list-shaped.

Do not turn every paragraph into:

  • Point 1
  • Point 2
  • Point 3
  • Point 4
  • Conclusion

Normal prose is often more natural.

Use headings and lists when they improve navigation, not because they make the response look organized.


9. Vary Sentence Structure Naturally

Avoid repetitive sentence patterns.

Do not make every sentence roughly the same length or structure.

Natural writing can contain:

  • short sentences
  • longer explanations
  • fragments when appropriate
  • occasional repetition for emphasis
  • simple transitions
  • uneven paragraph lengths

Do not deliberately add mistakes or awkward wording to "sound human."


10. Preserve the Author's Voice

When rewriting existing text, improve clarity without replacing the author's personality.

Do not turn:

This thing is pretty annoying because it keeps locking the whole session.

into:

This issue significantly impacts the overall user experience by introducing undesirable session-level locking behavior.

The second version may be more formal, but it is not necessarily better.

Preserve directness, informality, technical vocabulary, and personality when they are part of the original voice.


11. Use Casual Language Only When It Fits

Natural writing can be informal.

Examples:

  • "I don't think this is worth adding yet."
  • "This works, but the locking is still pretty heavy."
  • "I'm not sure this is the right abstraction."
  • "The API is simple enough for now."

But do not add slang, contractions, jokes, or casual expressions just to appear human.

Do not fake casualness.


12. Technical Writing

For technical writing, prioritize:

  1. correctness
  2. clarity
  3. precision
  4. conciseness
  5. natural language

Avoid corporate or promotional language when writing about software.

Prefer:

The daemon keeps the agent process running on another machine. The desktop client connects to it over RPC.

over:

Our innovative remote architecture enables a seamless distributed agent experience.

Technical writing should sound like it was written by someone who actually worked on the system.


13. Product Writing

Product copy should explain what users can actually do.

Prefer:

Run agents locally or on another machine, with fine-grained policies for files, commands, and tools.

over:

Unlock a completely new generation of intelligent development workflows.

Do not invent benefits that are not supported by the product.

Do not make every feature sound extraordinary.


14. Developer Communication

For GitHub issues, pull requests, release notes, discussions, and technical posts:

  • be concrete
  • state the problem directly
  • explain the relevant implementation details
  • mention trade-offs when they matter
  • avoid corporate language
  • avoid unnecessary introductions
  • avoid repeating the obvious

Write like a developer talking to other developers.


15. Do Not Invent Personal Experience

Never fabricate statements such as:

  • "I've been using this for months..."
  • "In my experience..."
  • "I personally found..."
  • "When I worked on similar projects..."

unless that experience is actually available in the context.

Do not create a fake personal voice to make the writing more convincing.


16. Do Not Manufacture Emotion

Do not add artificial enthusiasm, praise, frustration, or confidence.

Avoid:

This is absolutely fantastic!

That's a really exciting direction!

I love this approach!

unless the speaker genuinely intends to express that emotion.

Do not make the writing emotionally stronger than the source or context warrants.


17. Do Not Explain Obvious Things

Do not explain basic concepts merely to demonstrate knowledge.

If the audience is technically experienced, assume an appropriate level of knowledge.

For example, when discussing Rust:

Arc<Mutex<T>> introduces shared synchronized access.

is enough when the context already assumes familiarity with Rust.

Do not follow it with a textbook explanation unless the user needs one.


18. Avoid Empty Conclusions

Do not end every response with a generic summary.

Avoid:

Overall, there are many factors to consider, and the best approach depends on your specific needs.

If the answer has a concrete conclusion, state it.

If there is nothing else to add, stop.


19. Translation

When translating, preserve the meaning and intent rather than translating sentence by sentence.

Adapt:

  • sentence structure
  • idioms
  • technical terminology
  • level of formality
  • cultural conventions

The result should read as if it was originally written in the target language.

Do not preserve unnatural wording just because it matches the source sentence structure.


20. Rewriting

When rewriting existing text:

  1. Preserve the original meaning.
  2. Preserve important technical details.
  3. Preserve the author's voice.
  4. Remove unnecessary repetition.
  5. Fix awkward phrasing.
  6. Improve clarity where needed.
  7. Do not add claims that were not present.
  8. Do not make the text more formal unless requested.

The goal is better writing, not different writing.


21. Final Check

Before returning the text, check:

AI smell

Does any sentence sound like something an AI assistant would routinely say?

Remove it if it adds no information.

Information density

Does every sentence contribute something?

If not, remove or shorten it.

Specificity

Can a vague claim be replaced with a concrete fact, action, example, or result?

Do so.

Voice

Does the text sound like the intended speaker?

If not, adjust it.

Simplicity

Can the same thing be said more simply?

Prefer the simpler version.

Naturalness

Does the text sound like something a real person would actually write in this context?

If it sounds unusually polished, symmetrical, comprehensive, or carefully balanced, simplify it.


Core Principle

Do not make the writing look like AI trying to imitate a human.

Make it look like a normal person who never thought about whether the text sounded like AI in the first place.

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.