humanize
Skill1 starsChanged 2 months agoMITView on GitHub
Use when the user asks to make writing sound more natural, direct, human, or less "corporate"/"robotic"/"generic" — or asks why a draft feels stiff, bloated, or formulaic. Also useful proactively when generating new prose (emails, posts, docs, reports) in a register where filler phrases, hedge words, uniform sentence rhythm, or clichéd transitions would weaken it. This is a style and editing skill, not a detection or scoring tool.
78 other projects use a version of this.
Adopted by
Most recently taken up by
- fmind/dotfilestheir own version, first seen Nov 16, 2026
- fmind/dottheir own version, first seen Nov 15, 2026
- aya-taleb/instagram-boost-skillstheir own version, first seen Nov 15, 2026
- aya-taleb/entrepreneur-skillstheir own version, first seen Nov 15, 2026
- pivoshenko/pivoshenko.aitheir own version, first seen Nov 12, 2026
- pedrohcgs/claude-code-my-workflowtheir own version, first seen Nov 8, 2026
Used with
Projects with it also use humanize-redo, humanize-korean and 10 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.
dawn840705Skillhumanize-redo
가장 최근 윤문 결과를 2차로 다시 다듬는다 — 특정 카테고리·문단·강도 조정도 가능. humanize-korean 의 윤문 콜을 기존 run_id 에 재실행해 잔존 finding 을 처리한다. 트리거 — "$humanize-redo". **명시적 슬래시 호출 전용** — 자연어 후속 요청은 `humanize-korean` 오케스트레이터가 받는다.
Also in 5 projects that use this
mskim8717Skillhumanize-korean
AI(ChatGPT·Claude·Gemini 등)가 쓴 한글 텍스트를 "사람이 쓴 글처럼" 윤문해주는 오케스트레이터 스킬. 번역투·영어 인용 과다·기계적 병렬·관용구·피동태 남용·접속사 남발·리듬 균일성·이모지/불릿 과다 등 10대 카테고리 40+ AI 티 패턴을 탐지·분류해 내용은 한 글자도 건드리지 않고 문체·리듬·표현만 자연스러운 한국어로 재작성한다. 트리거 — "AI 티 없애줘", "AI 같은 글 자연스럽게", "GPT/ChatGPT 문체", "AI 번역투 고쳐", "사람이 쓴 것처럼 윤문", "AI 윤문", "ChatGPT 티 제거", "한글 AI 탐지·윤문", "AI 글 사람처럼", "번역투 제거", "영어 인용 많은 글 윤문", "AI 글 티 안 나게", "휴머나이저", "humanize Korean", "AI detector bypass 한글". 후속 작업 — "특정 카테고리만 다시", "윤문 강도 조정", "장르 바꿔서", "이 문단만", "2차 윤문" 도 모두 이 스킬. 단순 맞춤법·오탈자 교정은 직접 처리, 번역은 번역 스킬, 내용 추가·삭제를 동반한 재작성은 별도 집필 스킬.
Also in 3 projects that use this
CoralGameSkillperf-profile
Structured performance profiling workflow. Identifies bottlenecks, measures against budgets, and generates optimization recommendations with priority rankings.
Also in 3 projects that use this
Rayan55050Skillcontent
Создание контента: посты, статьи, тексты
Also in 3 projects that use this
luischanciSkillproofread
Run the proofreading protocol on manuscript files. Checks grammar, typos, overflow, consistency, and academic writing quality. Produces a report without editing files.
Also in 3 projects that use this
insight68Skillpricing
Help figure out pricing for a product or service using minimalist entrepreneur principles. Use when someone is setting prices, considering price changes, or struggling with what to charge.
Also in 3 projects that use this
Other versions78 projects use a version of this.
Word for word · 0
None found.
Edited · 0
None found.
Wrote their own version · 78
- epoko77-ai/im-not-ai★ 6.6ksame name
- pedrohcgs/claude-code-my-workflow★ 1.7ksame name
- PolyArch/humanize★ 1.5ksame name
- cwinvestments/memstack★ 433same name
- Mathews-Tom/armory★ 340same name
- whawkinsiv/solo-founder-skills★ 260same name
- whawkinsiv/claude-code-skills★ 258same name
- c-kick/hnl-agent-skills★ 56same name
- tzengyuxio/skills★ 45same name
- amondnet/yoonmoon★ 17same name
68 more not listed
How matches are found. Word for word: the exact same text. Edited: shares passages with this file (“99% same” means 99% of its five-word phrases also appear here; 10% or more counts). Their own version: similar meaning, or, weakest, the same name. We can’t tell which project had it first. Projects are counted among the 200 most-starred candidates, and mentions are links to the repository we found, up to 50 per site, so both counts are a floor.
In this repository10 other files it gives its agents.
- .agents/AGENTS.mdAGENTS.md
- AGENTS.mdAGENTS.md
- CLAUDE.mdCLAUDE.md
- .github/copilot-instructions.mdCopilot instructions
- ai-digestSkill
- daily-digestSkill
- linkedinSkill
- meSkill
- newsletterSkill
- pixelSkill
The file
Show its rules and text
We split this skill into the rules it gives an agent
16 rules. 0 are common, 13 are distinctive, and 3 are very specific to this project.
0
Common
Most of the 68 edited versions we read keep these.
13
Distinctive
This author's own choices. Worth a look.
3
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Very specific: line 29
Create this checklist first.
- Very specific: line 43
Load your profile & memory: Read config.profile (name, role, voice, brand, content pillars), the voice samples in content.md when tone …
- Distinctive: line 44
Ask if anything is missing: If a required input is unknown, stop and ask a specific question.
- Very specific: line 46
Validate, output & record: Check the result against this skill's rules, return it, then append any new durable fact to memory.md and any …
- Distinctive: line 51
Cut corporate/AI-cliché vocabulary: Use plain, specific words instead.
- Distinctive: line 120
Every few sentences, drop in a short fragment.
- Distinctive: line 121
Don't let three consecutive sentences land within 5 words of each other in length.
- Distinctive: line 127
Every abstract claim benefits from a grounding anchor — a number, a name, a date, a concrete example.
- Distinctive: line 137
Not every idea needs a bulleted list or a numbered section.
- Distinctive: line 181
If the draft reads like a chatbot answering a question rather than a person making a point, look for:
- Distinctive: line 199
When rewriting a draft, check in this order:
- Distinctive: line 200
Read it once for content — does it say what it needs to say?
- Distinctive: line 202
Check sentence-length variation (section 3) — split or merge as needed.
- Distinctive: line 203
Check every abstract claim has a concrete anchor, or an honest approximation (section 4).
- Distinctive: line 204
Check for unnecessary bullet/numbered structure (section 5) — convert to prose where it reads better that way.
- Distinctive: line 206
Read it once more for voice consistency and cut any assistant-voice tics (section 8).
“Common” is measured: kept by at least 6 in 10 of the 68 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.
Files
Natural Writing Style
Role
You are a writing editor. You rewrite stiff, bloated, or generic prose into direct, specific, natural-sounding writing.
Context to load
Load and honor these before acting:
config.profile— identity, role, voice, brand, content pillars (personalize everything)content.md— voice samples and prompt guidance (when tone matters)memory.md/notes.md— learned preferences and open follow-ups
Workflow
Step 0: TodoWrite Checklist
Create this checklist first; mark one task in_progress, finish it, then move to the next.
TodoWrite([
{ content: "Gather inputs; load config.profile, content.md, memory.md & notes.md", status: "in_progress" },
{ content: "Ask for any missing inputs — never assume or hallucinate", status: "pending" },
{ content: "Rewrite the supplied text to read naturally", status: "pending" },
{ content: "Validate output; record new facts to memory.md / follow-ups to notes.md", status: "pending" }
])Steps
- Gather inputs. Parse the request /
$ARGUMENTSand list what this skill needs. - Load your profile & memory. Read
config.profile(name, role, voice, brand, content pillars), the voice samples incontent.mdwhen tone matters, and the sharedmemory.md/notes.mdfor learned preferences and open follow-ups. - Ask if anything is missing. If a required input is unknown, stop and ask a specific question. Never assume, guess, or invent facts, numbers, names, or sources.
- Do the work following the sections below.
- Validate, output & record. Check the result against this skill's rules, return it, then append any new durable fact to
memory.mdand any follow-up tonotes.mdviapython .agents/scripts/memory.py.
The core moves
- Cut hedge words and filler wrappers. Say the thing directly.
- Cut corporate/AI-cliché vocabulary. Use plain, specific words instead.
- Vary sentence length on purpose. Uniform medium-length sentences read flat.
- Replace vague claims with concrete anchors. Numbers, names, dates, examples.
- Let structure emerge from content, rather than imposing bullet lists and subheadings on everything.
- Match voice to register — a Slack update, a memo, and an essay don't sound alike.
- Drop assistant-voice tics — "helpful assistant" framing reads as generic in any register, human or AI-drafted.
1. Hedge and filler surgery
Delete or replace:
- "it is important to note that", "it is worth mentioning that", "generally speaking", "in many cases", "it can be argued", "often" / "typically" (unless genuinely needed for accuracy)
- Announcement-colon openers: "The rule I use:", "The key insight:" — just state it.
Filler-wrapper substitutions (the pattern generalizes — any multi-word wrapper around a one-word meaning gets the one word):
| Verbose | Direct |
| Due to the fact that | Because |
| In the event that | If |
| Has the ability / capacity to | Can |
| Make a decision / an assumption | Decide / Assume |
| For the purpose of | To / For |
| With regard to / With respect to | About / On |
| Prior to / Subsequent to | Before / After |
| In light of the fact that / Despite the fact that | Since / Although |
Real uncertainty gets human phrasing: "I'm not sure this holds for edge cases" beats "while results may vary."
2. Vocabulary to avoid (and what to use instead)
These words aren't wrong, but they're overused to the point of sounding like filler. Swap for a plainer, more specific word chosen for the actual context.
- Verbs: delve, leverage, utilize, streamline, foster, facilitate, garner, showcase, highlight (as standalone verb), underscore (as standalone verb) → pick the concrete verb: "use", "help", "cut", "show"
- Adjectives: robust, comprehensive, pivotal, nuanced, multifaceted, crucial (overused), enduring, vibrant, intricate → say what specifically makes it that
- Abstract nouns: landscape (figurative), tapestry (figurative), testament (figurative), interplay, intricacies → name the actual thing
- Openers/closers: "in today's fast-paced world", "in conclusion", "in summary", "it goes without saying", "at the end of the day" → cut entirely
- Transitions: furthermore, moreover, "it is clear that", "this highlights", "this underscores" → cut, or use "and"/"but"/"so"
- Significance inflation: "stands as a testament to", "marks a pivotal moment in", "evolving landscape", "plays a vital role" → state the concrete fact instead
- Promotional register: "nestled in the heart of", "boasts a rich heritage", "breathtaking", "must-visit", "renowned for" → cut the brochure language
- Quantifier inflation: "a myriad of", "a plethora of", "in the realm of"
- Tutorial scaffolding: "let's dive in", "let's break this down", "without further ado" → just start
- Sycophantic prefixes: "great question", "you're absolutely right", "of course!"
- Templated closers: "happy to jump on a call", "feel free to reach out", "I hope this helps"
3. Sentence rhythm
Uniform sentence length (most sentences landing in the same 10–20 word band) reads monotonous. Aim for real variation:
- Mix short, punchy sentences with longer ones that build across a clause or two.
- Every few sentences, drop in a short fragment. Like that.
- Don't let three consecutive sentences land within 5 words of each other in length.
- A run of only short sentences reads choppy; a run of only long ones reads exhausting. Both directions need a counterweight.
4. Specificity over abstraction
Every abstract claim benefits from a grounding anchor — a number, a name, a date, a concrete example.
"Performance improved significantly" → "Latency dropped from 340ms to 80ms under the same load." If exact specifics aren't available, use honest approximation framing: "in the cases I've seen...", "the one time this bit us..." — never invent numbers or examples just to sound specific.
5. Structural flattening
Not every idea needs a bulleted list or a numbered section. Some patterns to reconsider:
| Formulaic pattern | More natural alternative |
| Intro sentence + 3-bullet list | A prose paragraph where the items connect through flow |
| "There are three main factors: ..." | Just talk about the factors; let transitions carry it |
| Numbered sections for everything | Sections only when order genuinely matters |
| Topic sentence + evidence + restatement | Skip the restatement — don't recap what you just said |
| Vague attributions: "industry observers note" | Name a specific source, or drop the claim |
| "Challenges and Future Prospects" boilerplate section | The actual challenges, specifically, or cut the section |
6. Voice and register
Writing carries a perspective. Depending on register:
- First person where natural: "I find that...", "In my experience..."
- Occasional direct address: "If you've run into this before..."
- Contractions in conversational contexts: "don't", "it's", "you'll"
- Self-correction mid-thought when it's genuine: "— actually, more precisely:"
Calibrate to domain:
- Technical: domain-native vocabulary ("the hot path," "this falls apart at scale"), short sentences for definitive claims, direct tradeoffs, real tool/version names.
- Narrative/essay: open with a scene or specific moment rather than a thesis statement; let the argument emerge from the evidence.
- Professional/business: cut the throat-clearing opener, state the ask in the first line or two, short paragraphs (2–3 sentences in email/memo).
- Slack/async updates: fragments are normal, thoughts loop and self-correct,
approximations over precise figures (
~60%,<10min), lowercase is fine outside proper nouns.
7. Punctuation habits worth checking
None of these are "wrong" — they're just easy to overuse without noticing:
- Em dashes: fine occasionally for a genuine interruption; if every third sentence has one, most will read cleaner as a period, comma, or restructure.
- Semicolons: rare outside formal/academic registers. A period, or "and"/"but"/"so," usually reads more naturally.
- Mid-sentence colons: clean at the end of a complete clause introducing something; awkward mid-thought ("The problem: nobody tests this" → "Nobody tests this.").
- Curly vs. straight quotes: match whatever your target publishing context expects.
8. Cut assistant-voice framing
If the draft reads like a chatbot answering a question rather than a person making a point, look for:
| Tell | Fix |
| "Here's how I'd think about it...", "Let me walk you through..." | Cut the framing, say the thing |
| "On one hand X, on the other Y, it depends..." | Pick a side; the reader can disagree |
| Unrequested enumeration of options | Answer directly; note the constraint after, if needed |
| Defining terms the audience already knows | Cut — trust the reader |
| A caveat appended to every claim | State the claim; caveat only real edge cases |
| "That's a great question, and..." | Cut entirely |
| Closing summary recapping what was just said | Cut |
| "I hope this helps! Let me know if..." | End on the last substantive sentence |
Quick editing pass
When rewriting a draft, check in this order:
- Read it once for content — does it say what it needs to say?
- Scan for banned vocabulary and filler wrappers (section 2) — replace or cut.
- Check sentence-length variation (section 3) — split or merge as needed.
- Check every abstract claim has a concrete anchor, or an honest approximation (section 4).
- Check for unnecessary bullet/numbered structure (section 5) — convert to prose where it reads better that way.
- Read it once more for voice consistency and cut any assistant-voice tics (section 8).
Output the rewritten text itself — skip the "Here's the revised version:" preamble and the "what I changed" recap unless the user asks for one.
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
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.

