agentleFS
eherrerosjeherrerosj/claude-code-toolkit/skills/humanize/SKILL.md

humanize

Skill1 starsChanged 8 months agoMITView on GitHub

Remove signs of AI-generated writing from text. Use when editing or reviewing text to make it sound more natural and human-written. Based on Wikipedia's comprehensive 'Signs of AI writing' guide. Detects and fixes patterns including: inflated symbolism, promotional language, superficial -ing analyses, vague attributions, em dash overuse, rule of three, AI vocabulary words, negative parallelisms, and excessive conjunctive phrases. Triggers on: 'humanize', 'make this sound human', 'remove AI writing', 'de-AI', 'sound more natural', 'less robotic', 'too AI-sounding', or when reviewing text that reads like AI output.

We split this skill into the rules it gives an agent

7 rules. 0 are common, 7 are distinctive, and 0 are very specific to this project.

  • 0

    Common

    Most of the 56 edited versions we read keep these.

  • 7

    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:

    Read the full text

    line 16
  2. Distinctive:

    Fix: Use concrete, specific descriptions.

    line 67
  3. Distinctive:

    Keep em dashes only where they genuinely improve rhythm

    line 131
  4. Distinctive:

    Detection: Look for three-item lists, three adjectives in a row, three parallel clauses.

    line 144
  5. Distinctive:

    Fix: Use the number of items the content actually demands.

    line 146
  6. Distinctive:

    Fix: Use the simplest accurate word.

    line 167
  7. Distinctive:

    Fix: Use formatting sparingly. Bold only genuinely critical terms.

    line 240

“Common” is measured: kept by at least 6 in 10 of the 56 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

Humanize

Strip AI fingerprints from text. Make writing sound like a specific human wrote it, not a statistical average of all humans.

The Core Problem

AI doesn't write badly — it writes generically. It produces the statistically most likely phrasing, which sounds polished but hollow. Humanizing isn't just removing bad patterns; it's adding authentic voice: varied rhythm, specific details, genuine opinions, and the occasional rough edge that proves a real person wrote it.

Process

  1. Read the full text
  2. Scan for patterns from each category below
  3. Rewrite flagged passages — don't just delete; replace with something specific and natural
  4. Self-audit: read the result and ask "What still screams AI?" Fix those
  5. Final pass: vary sentence length, break any remaining rhythmic monotony

Pattern Catalog

1. Inflated Symbolism & Significance

AI exaggerates importance of mundane facts. Everything becomes pivotal, transformative, a testament.

Kill phrases:

  • "stands as / serves as / is a testament to"
  • "plays a vital/significant/crucial role"
  • "underscores its importance"
  • "watershed moment", "key turning point"
  • "continues to captivate / resonate / inspire"
  • "leaves a lasting impact / legacy"
  • "deeply rooted", "profound heritage"
  • "steadfast dedication", "solidifies"
  • "marks a crucial phase", "a pivotal step"

Fix: State the fact plainly. If something is genuinely significant, show it through specifics and evidence — don't announce it.

BAD:  "The 2019 redesign stands as a testament to the company's enduring commitment to innovation."
GOOD: "The 2019 redesign replaced the dashboard with a single-page app, cutting load times by 40%."

2. Promotional Language

Travel-brochure tone. Everything is stunning, breathtaking, rich.

Kill phrases:

  • "rich cultural heritage / tapestry"
  • "breathtaking views / scenery"
  • "stunning natural beauty"
  • "must-visit / must-see"
  • "enduring / lasting legacy"
  • "nestled in the heart of"
  • "boasts a range of"
  • "offers a glimpse into"
  • "clean and modern facilities"
  • "vibrant community / atmosphere"
  • "cutting-edge", "world-class"

Fix: Use concrete, specific descriptions. Replace adjectives with facts.

BAD:  "Nestled in the heart of the valley, this breathtaking destination boasts stunning natural beauty."
GOOD: "The town sits at the valley floor, surrounded by 3,000-metre peaks on three sides."

3. Superficial -ing Analysis

Trailing participial clauses that gesture at meaning without delivering it. The sentence ends with "...highlighting / emphasizing / reflecting / ensuring / showcasing" followed by a vague claim.

Kill patterns:

  • "...highlighting the importance of X"
  • "...emphasizing the need for Y"
  • "...reflecting the continued relevance of Z"
  • "...ensuring a brighter future"
  • "...showcasing the diversity of"
  • "...underscoring the significance of"
  • "...demonstrating a commitment to"

Fix: Either make a specific claim with evidence, or cut the trailing clause entirely — the sentence is usually complete without it.

BAD:  "The policy was revised in 2022, reflecting the government's commitment to equity."
GOOD: "The policy was revised in 2022 to extend parental leave from 12 to 16 weeks."

4. Vague Attribution (Weasel Words)

Attributing claims to unnamed authorities. Inflating how widely a view is held.

Kill patterns:

  • "Industry reports suggest..."
  • "Observers have cited..."
  • "Some critics argue..."
  • "Many scholars believe..."
  • "Experts widely agree..."
  • "It is generally considered..."
  • Mentioning multiple "reviewers" or "scholars" while citing only one
  • Implying a non-exhaustive list ("among others") without evidence

Fix: Name the source, or drop the claim. If you can't attribute it, it probably doesn't belong.

BAD:  "Many experts consider this approach revolutionary."
GOOD: "Chen (2021) called this approach 'the first practical alternative to gradient descent.'"

5. Em Dash Overuse

AI uses em dashes where humans would use commas, parentheses, colons, or just a period and new sentence. One or two per piece is fine — five is a tell.

Detection: Count em dashes (—). More than 2 per 500 words is suspicious.

Fix: Replace most with the punctuation a human would naturally choose:

  • Parenthetical aside → use parentheses or commas
  • Introducing an explanation → use a colon
  • Dramatic pause → just start a new sentence
  • Keep em dashes only where they genuinely improve rhythm
BAD:  "The framework — originally designed for mobile — was adapted for desktop — a move that surprised many — and quickly gained traction."
GOOD: "The framework was originally designed for mobile. Its adaptation for desktop surprised many, but it quickly gained traction."

6. Rule of Three

AI defaults to triplets when listing anything: adjectives, benefits, examples, takeaways. Occasionally natural, but AI does it reflexively.

Detection: Look for three-item lists, three adjectives in a row, three parallel clauses. Especially suspicious when the third item adds nothing beyond the first two.

Fix: Use the number of items the content actually demands. Sometimes that's one. Sometimes it's four. Rarely is it exactly three.

BAD:  "The platform offers speed, reliability, and scalability."
GOOD: "The platform is fast — it handles 10k requests per second on commodity hardware."

7. AI Vocabulary Words

Specific words that appear at far higher rates in AI text than in human writing. These shift with model generations.

High-frequency AI words (always suspect): delve, tapestry, landscape, pivotal, underscore, crucial, intricate/intricacies, meticulous/meticulously, testament, vibrant, foster, bolster, beacon, realm, nuanced, multifaceted, interplay, paramount, embark, encompass, leverage, harness, illuminate, facilitate, garner, endeavor

Era-specific tells:

  • GPT-4 era (2023–mid 2024): delve, underscore, intricate, tapestry, meticulous
  • GPT-4o era (mid 2024–mid 2025): emphasizing, enhance, highlighting, showcasing
  • Later models: these shift but the pattern of overuse persists

Fix: Use the simplest accurate word. "Important" instead of "pivotal." "Complex" instead of "multifaceted." "Explore" instead of "delve into." Or better yet, be specific instead of reaching for any adjective.

BAD:  "This nuanced, multifaceted approach delves into the intricate interplay of factors."
GOOD: "This approach examines how cost, adoption rate, and regulatory pressure interact."

8. Negative Parallelism

"Not X, but Y" constructions used excessively. Especially "It's not just about X — it's about Y."

Kill patterns:

  • "It's not just about X, it's about Y"
  • "Not only... but also..."
  • "It's not X — it's Y"
  • "No X, no Y — just Z"
  • "More than just X, it represents Y"

Fix: Just state what it IS. The negation frame rarely adds value.

BAD:  "It's not just about the technology — it's about the people behind it."
GOOD: "The team matters more than the stack."

9. Excessive Conjunctive Phrases

Overuse of transition words suggesting analytical depth that isn't there.

Overused transitions:

  • "Moreover", "Furthermore", "Additionally"
  • "However", "Nevertheless", "Nonetheless"
  • "In contrast", "On the other hand"
  • "It is worth noting", "It is important to note"
  • "In summary", "Overall"
  • "No discussion would be complete without"

Fix: Cut most of them. Good writing flows without signposts every sentence. When a transition is truly needed, prefer shorter ones ("But", "Still", "Also") or restructure so the logic is implicit.

BAD:  "Furthermore, the system provides real-time monitoring. Moreover, it supports alerting. Additionally, dashboards are customizable."
GOOD: "The system monitors in real time, fires alerts on threshold breaches, and lets you build custom dashboards."

10. Editorializing & Sycophantic Tone

Inserting opinions disguised as objectivity. Excessive positivity. Complimenting the reader.

Kill patterns:

  • "It's important to note/remember/consider"
  • "Great question!" / "That's a really interesting point"
  • "This is a fascinating area"
  • "In this article, we will explore..."
  • Any sentence starting with "Certainly" or "Absolutely"

Fix: Drop the editorial frame. Present information, not your feelings about the information.


11. Formatting Tells

  • Excessive boldface — bolding every key term
  • Inline-header lists — bold phrase followed by colon, then explanation restating the bold phrase
  • Bulleted lists where prose would be more natural
  • Title Case On Every Heading Word
  • Consistent use of curly quotes when context suggests straight quotes

Fix: Use formatting sparingly. Bold only genuinely critical terms. Prefer prose over lists when items are closely related. Use sentence case for headings.


12. Copula Avoidance & Synonym Cycling

AI avoids repeating "is" by substituting elaborate constructions ("represents", "constitutes", "embodies"). It also cycles through synonyms for the same noun to avoid repetition, creating confusion.

Kill patterns:

  • "represents a significant advancement"
  • "constitutes a fundamental shift"
  • Referring to the same thing as "the platform", "the solution", "the offering", "the tool" in consecutive sentences

Fix: Repeat "is" and repeat nouns. Clarity beats variety. Call the same thing the same name throughout.


13. False Ranges & Compulsive Summaries

"From X to Y" implying a spectrum between unrelated things. Reflexive "Overall" or "In conclusion" paragraphs even in short text.

Kill patterns:

  • "From X to Y" (where X and Y aren't ends of a real spectrum)
  • "Overall, ..." as final paragraph opener
  • "In conclusion, ..." in anything under 1000 words
  • Restating everything just said

Fix: Cut the false range. Cut the summary if the piece is short enough to not need one. If a summary is warranted, add new insight — don't restate.


Self-Audit Checklist

After rewriting, verify:

  • No more than 1 em dash per 300 words
  • No triplet lists that could be 1, 2, or 4 items
  • Zero words from the AI vocabulary list (or a deliberate, justified use)
  • No trailing "-ing" clauses making vague significance claims
  • No unnamed experts, scholars, or observers
  • Sentence lengths vary (mix of short, medium, long)
  • At least one opinion, specific detail, or rough edge that a generic AI wouldn't produce
  • No "not X but Y" constructions unless genuinely needed for contrast
  • Formatting (bold, lists, headings) used sparingly and purposefully
  • Read it aloud — does it sound like a specific person, or like a press release?

Download now (ZIP)

  • Other versions66 projects use a version of this.

    Word for word · 0

    None found.

    Edited · 0

    None found.

    Wrote their own version · 66

    1. epoko77-ai/im-not-ai★ 6ksame name
    2. pedrohcgs/claude-code-my-workflow★ 1.6ksame name
    3. PolyArch/humanize★ 1.5ksame name
    4. cwinvestments/memstack★ 425same name
    5. whawkinsiv/solo-founder-skills★ 260same name
    6. whawkinsiv/claude-code-skills★ 258same name
    7. 3x-haust/oh-my-design★ 203same name
    8. c-kick/hnl-agent-skills★ 56same name
    9. tzengyuxio/skills★ 45same name
    10. amondnet/yoonmoon★ 17same name

    56 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.

  • Used alongsideProjects 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.
    • dawn840705Skill

      humanize-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

    • mskim8717Skill

      humanize-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

    • luischanciSkill

      proofread

      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

    • CoralGameSkill

      perf-profile

      Structured performance profiling workflow. Identifies bottlenecks, measures against budgets, and generates optimization recommendations with priority rankings.

      Also in 3 projects that use this

    • Rayan55050Skill

      content

      Создание контента: посты, статьи, тексты

      Also in 3 projects that use this

    • slavingiaSkill

      pricing

      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

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

Reviews can't be read right now.

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.