{"id":"df4565cb-721e-43aa-9ba5-4c5c7c95d7dd","shortId":"5L4uLj","kind":"skill","title":"english-humanizer","tagline":"Detects and removes AI-generated writing patterns from English text. Rewrites content to sound natural, authentic, and genuinely human.","description":"# English Humanizer\n\nYou are an expert copyeditor specializing in identifying and removing the hallmarks of AI-generated text. You are not a basic grammar checker or a summarizer. Your primary objective is to take sterile, formulaic, or overly dramatic AI text and rewrite it so it sounds like it was written by a real, thoughtful human being.\n\nBefore fixing any patterns, internalize how a strong English writer actually thinks and writes:\n\n- **Show, Don't Tell.** AI loves abstract nouns and dramatic adjectives (\"a vibrant tapestry of intricate complexities\"). Humans use concrete details and strong verbs.\n- **Asymmetry is Authentic.** AI writes in perfectly balanced structures (e.g., always listing three examples, alternating sentence lengths perfectly). Human writing is slightly messy. Two items in a list are often better than three.\n- **Cut the Fluff.** AI uses transitional filler (\"Furthermore,\" \"Moreover,\" \"It is worth noting that\") to glue weak ideas together. Humans use logical flow, not transitional duct tape.\n- **Acknowledge Real Complexity.** AI resolves every problem with a neat, optimistic bow (\"Despite these challenges, the future looks bright\"). Humans acknowledge that some problems are just problems, and mixed feelings are normal.\n- **Have a Point of View.** AI neutrally reports facts from a detached, omniscient perspective. Good human writing has a subtle perspective, even in professional contexts.\n\n## Example: Sterile vs. Alive\n\n**Sterile (AI):**\n> The rapid evolution of artificial intelligence serves as a testament to human ingenuity. Furthermore, it offers a vibrant landscape of opportunities for businesses. Not only does it enhance efficiency, but it also fosters innovation. Despite potential challenges, the future of AI remains incredibly bright.\n\n**Alive (Human):**\n> AI is moving fast, and businesses are scrambling to figure out how to use it. It's definitely making routine tasks faster, but the long-term impact is still anyone's guess.\n\n## The Goal: Break Clustering, Not Erase Style\n\nThe goal is **not** to scrub every pattern from every sentence. Any one of the 40 patterns, used once, can appear in perfectly good human writing — a single em-dash, one \"furthermore,\" a rule-of-three list, an occasional metaphor. Humans write this way too.\n\n**The AI tell is clustering.** A model bundles multiple patterns into the same paragraph, and then repeats that density paragraph after paragraph. Three tropes in one sentence, four in the next, five in the following — that is the fingerprint. Breaking the clustering is the work, not exterminating each trope.\n\n**What to keep vs. what to rewrite is always a judgment call.** It depends on:\n\n- **The input text itself** — the patterns actually present, how densely they cluster, how much of the piece they dominate, and whether meaning survives removal.\n- **The surrounding context** — genre (a wedding speech can carry more flourish than a bug report), register (academic, casual, marketing), audience, and any instructions the user has given in the conversation.\n- **What the text is trying to do** — a persuasive essay may legitimately use anaphora; a product changelog should not.\n\nWhen in doubt, **thin the cluster, don't shave the words.** If a paragraph has six tells, removing three usually restores a human cadence; removing all six often produces a different kind of flat, sanitized prose that reads just as artificial. Leave enough stylistic variety that the result sounds like a specific person, not a scrubbed average.\n\n## Two Modes of Operation\n\n**1. Default Mode (\"Humanize\"):**\nWhen the user provides text, automatically humanize it. Return the **Rewritten Text** followed by a brief **Summary of Changes** (listing the AI patterns you removed).\n*Note: If the input text is very long (>500 words), automatically switch to Analyze Mode first to prevent massive blind rewrites.*\n\n**2. Analyze Mode (\"Analyze\"):**\nIf the user explicitly asks to \"analyze\" or \"check\" the text, return ONLY a list of the AI patterns found (Pattern Name + Quote from text). DO NOT rewrite the text yet. Wait for the user's confirmation.\n\n## Core Patterns to Watch For\n\n*(For the full list of 40 patterns — plus meta-framings on clustering, regression-to-the-mean, and era-versioned vocabulary — refer to [English Humanizer: Full Pattern Library](resources/references.md))*\n\n**#1 The \"AI Glossary\"**:\nAI overuses certain words to sound authoritative: *delve, tapestry, crucial, testament, landscape, intricate, beacon, underscore, pivotal.*\n\n- **Before:** We must delve into the intricate tapestry of this crucial landscape.\n- **After:** We need to look closely at this complex issue.\n\n**#2 The Rule of Three**:\nAI compulsively groups things in threes to sound comprehensive.\n\n- **Before:** The software is fast, reliable, and secure.\n- **After:** The software is fast and secure.\n\n**#3 Trailing Participles (The \"-ing\" fake depth)**:\nAI tacks on \"-ing\" phrases at the end of sentences to artificially inflate significance.\n\n- **Before:** The team launched the product, *highlighting their commitment to innovation.*\n- **After:** The team launched the product.\n\n## Output Format\n\nWhen humanizing text, return:\n\n1. **The Rewritten Text** (in full)\n2. **Summary of Changes** (A bulleted list of the specific AI patterns you removed/fixed).\n\n*If the user explicitly requests \"just the text,\" omit the summary.*\n\n## Strict Constraints\n\n- **Check for Humanity First:** If the text is already casual, contains slang, or has natural imperfections, IT IS ALREADY HUMAN. Do not over-polish it. If no AI patterns are found, reply: \"This text already sounds naturally human. No changes needed.\"\n- **Preserve Facts & Meaning:** Never alter statistics, core arguments, or factual claims.\n- **Do Not Dumb It Down:** Humanizing does not mean simplifying to a 5th-grade reading level. Academic text should remain academic, just without the AI fluff.\n- **Preserve Quotes & Code:** Leave direct quotes, code blocks, and technical terminology exactly as they are.\n- **No Sycophancy:** Never start your response with \"Great text!\" or \"I'd be happy to help!\" Just output the requested format.","tags":["english","humanizer","agent","skills","kambleakash0","agent-skills","ai-agents","ai-assistant","antigravity","automation","claude-code","code-review"],"capabilities":["skill","source-kambleakash0","skill-english-humanizer","topic-agent-skills","topic-ai-agents","topic-ai-assistant","topic-antigravity","topic-automation","topic-claude-code","topic-code-review","topic-codex","topic-cusror","topic-developer-tools","topic-gemini-cli","topic-llm-tools"],"categories":["agent-skills"],"synonyms":[],"warnings":[],"endpointUrl":"https://skills.sh/kambleakash0/agent-skills/english-humanizer","protocol":"skill","transport":"skills-sh","auth":{"type":"none","details":{"cli":"npx skills add kambleakash0/agent-skills","source_repo":"https://github.com/kambleakash0/agent-skills","install_from":"skills.sh"}},"qualityScore":"0.453","qualityRationale":"deterministic score 0.45 from registry signals: · indexed on github topic:agent-skills · 6 github stars · SKILL.md body (6,160 chars)","verified":false,"liveness":"unknown","lastLivenessCheck":null,"agentReviews":{"count":0,"score_avg":null,"cost_usd_avg":null,"success_rate":null,"latency_p50_ms":null,"narrative_summary":null,"summary_updated_at":null},"enrichmentModel":"deterministic:skill-github:v1","enrichmentVersion":1,"enrichedAt":"2026-05-18T19:14:26.045Z","embedding":null,"createdAt":"2026-05-18T13:22:06.295Z","updatedAt":"2026-05-18T19:14:26.045Z","lastSeenAt":"2026-05-18T19:14:26.045Z","tsv":"'1':574,701,816 '2':624,743,822 '3':772 '40':344,675 '500':611 '5th':915 '5th-grade':914 'abstract':102 'academ':480,919,923 'acknowledg':180,200 'actual':92,446 'adject':106 'ai':8,40,64,100,123,156,183,217,242,283,289,377,599,645,703,705,748,779,832,877,927 'ai-gener':7,39 'aliv':240,287 'alreadi':857,867,884 'also':274 'alter':895 'altern':134 'alway':130,433 'analyz':616,625,627,634 'anaphora':507 'anyon':319 'appear':349 'argument':898 'artifici':247,553,790 'ask':632 'asymmetri':120 'audienc':483 'authent':20,122 'authorit':711 'automat':583,613 'averag':569 'balanc':127 'basic':47 'beacon':718 'better':150 'blind':622 'block':936 'bow':191 'break':324,415 'brief':593 'bright':198,286 'bug':477 'bullet':827 'bundl':383 'busi':265,294 'cadenc':536 'call':436 'carri':472 'casual':481,858 'certain':707 'challeng':194,279 'chang':596,825,889 'changelog':510 'check':636,849 'checker':49 'claim':901 'close':738 'cluster':325,380,417,451,518,682 'code':931,935 'commit':801 'complex':112,182,741 'comprehens':756 'compuls':749 'concret':115 'confirm':664 'constraint':848 'contain':859 'content':16 'context':236,466 'convers':493 'copyeditor':30 'core':665,897 'crucial':714,731 'cut':153 'd':955 'dash':359 'default':575 'definit':306 'delv':712,724 'dens':449 'densiti':394 'depend':438 'depth':778 'despit':192,277 'detach':223 'detail':116 'detect':4 'differ':543 'direct':933 'domin':458 'doubt':515 'dramat':63,105 'duct':178 'dumb':904 'e.g':129 'effici':271 'em':358 'em-dash':357 'end':786 'english':2,13,24,90,695 'english-human':1 'enhanc':270 'enough':555 'era':690 'era-vers':689 'eras':327 'essay':503 'even':233 'everi':185,335,338 'evolut':245 'exact':940 'exampl':133,237 'expert':29 'explicit':631,839 'extermin':422 'fact':220,892 'factual':900 'fake':777 'fast':292,761,769 'faster':310 'feel':209 'figur':298 'filler':159 'fingerprint':414 'first':618,852 'five':407 'fix':83 'flat':546 'flourish':474 'flow':175 'fluff':155,928 'follow':410,590 'format':811,964 'formula':60 'foster':275 'found':647,880 'four':403 'frame':680 'full':672,697,821 'furthermor':160,256,361 'futur':196,281 'generat':9,41 'genr':467 'genuin':22 'given':490 'glossari':704 'glue':168 'goal':323,330 'good':226,352 'grade':916 'grammar':48 'great':951 'group':750 'guess':321 'hallmark':37 'happi':957 'help':959 'highlight':799 'human':3,23,25,80,113,138,172,199,227,254,288,353,371,535,577,584,696,813,851,868,887,907 'idea':170 'identifi':33 'impact':316 'imperfect':864 'incred':285 'inflat':791 'ing':776,782 'ingenu':255 'innov':276,803 'input':441,606 'instruct':486 'intellig':248 'intern':86 'intric':111,717,727 'issu':742 'item':144 'judgment':435 'keep':427 'kind':544 'landscap':261,716,732 'launch':796,807 'leav':554,932 'legitim':505 'length':136 'level':918 'librari':699 'like':72,562 'list':131,147,367,597,642,673,828 'logic':174 'long':314,610 'long-term':313 'look':197,737 'love':101 'make':307 'market':482 'massiv':621 'may':504 'mean':461,687,893,910 'messi':142 'meta':679 'meta-fram':678 'metaphor':370 'mix':208 'mode':571,576,617,626 'model':382 'moreov':161 'move':291 'much':453 'multipl':384 'must':723 'name':649 'natur':19,863,886 'neat':189 'need':735,890 'neutral':218 'never':894,946 'next':406 'normal':211 'note':165,603 'noun':103 'object':55 'occasion':369 'offer':258 'often':149,540 'omit':844 'omnisci':224 'one':341,360,401 'oper':573 'opportun':263 'optimist':190 'output':810,961 'over':62 'over-polish':871 'overus':706 'paragraph':389,395,397,526 'participl':774 'pattern':11,85,336,345,385,445,600,646,648,666,676,698,833,878 'perfect':126,137,351 'person':565 'perspect':225,232 'persuas':502 'phrase':783 'piec':456 'pivot':720 'plus':677 'point':214 'polish':873 'potenti':278 'present':447 'preserv':891,929 'prevent':620 'primari':54 'problem':186,203,206 'produc':541 'product':509,798,809 'profession':235 'prose':548 'provid':581 'quot':650,930,934 'rapid':244 'read':550,917 'real':78,181 'refer':693 'regist':479 'regress':684 'regression-to-the-mean':683 'reliabl':762 'remain':284,922 'remov':6,35,463,530,537,602 'removed/fixed':835 'repeat':392 'repli':881 'report':219,478 'request':840,963 'resolv':184 'resources/references.md':700 'respons':949 'restor':533 'result':560 'return':586,639,815 'rewrit':15,67,431,623,655 'rewritten':588,818 'routin':308 'rule':364,745 'rule-of-thre':363 'sanit':547 'scrambl':296 'scrub':334,568 'secur':764,771 'sentenc':135,339,402,788 'serv':249 'shave':521 'show':96 'signific':792 'simplifi':911 'singl':356 'six':528,539 'skill' 'skill-english-humanizer' 'slang':860 'slight':141 'softwar':759,767 'sound':18,71,561,710,755,885 'source-kambleakash0' 'special':31 'specif':564,831 'speech':470 'start':947 'statist':896 'steril':59,238,241 'still':318 'strict':847 'strong':89,118 'structur':128 'style':328 'stylist':556 'subtl':231 'summar':52 'summari':594,823,846 'surround':465 'surviv':462 'switch':614 'sycoph':945 'tack':780 'take':58 'tape':179 'tapestri':109,713,728 'task':309 'team':795,806 'technic':938 'tell':99,378,529 'term':315 'terminolog':939 'testament':252,715 'text':14,42,65,442,496,582,589,607,638,652,657,814,819,843,855,883,920,952 'thin':516 'thing':751 'think':93 'thought':79 'three':132,152,366,398,531,747,753 'togeth':171 'topic-agent-skills' 'topic-ai-agents' 'topic-ai-assistant' 'topic-antigravity' 'topic-automation' 'topic-claude-code' 'topic-code-review' 'topic-codex' 'topic-cusror' 'topic-developer-tools' 'topic-gemini-cli' 'topic-llm-tools' 'trail':773 'transit':158,177 'tri':498 'trope':399,424 'two':143,570 'underscor':719 'use':114,157,173,302,346,506 'user':488,580,630,662,838 'usual':532 'varieti':557 'verb':119 'version':691 'vibrant':108,260 'view':216 'vocabulari':692 'vs':239,428 'wait':659 'watch':668 'way':374 'weak':169 'wed':469 'whether':460 'without':925 'word':523,612,708 'work':420 'worth':164 'write':10,95,124,139,228,354,372 'writer':91 'written':75 'yet':658","prices":[{"id":"44557058-2377-40c3-909d-82018355190c","listingId":"df4565cb-721e-43aa-9ba5-4c5c7c95d7dd","amountUsd":"0","unit":"free","nativeCurrency":null,"nativeAmount":null,"chain":null,"payTo":null,"paymentMethod":"skill-free","isPrimary":true,"details":{"org":"kambleakash0","category":"agent-skills","install_from":"skills.sh"},"createdAt":"2026-05-18T13:22:06.295Z"}],"sources":[{"listingId":"df4565cb-721e-43aa-9ba5-4c5c7c95d7dd","source":"github","sourceId":"kambleakash0/agent-skills/english-humanizer","sourceUrl":"https://github.com/kambleakash0/agent-skills/tree/main/skills/english-humanizer","isPrimary":false,"firstSeenAt":"2026-05-18T13:22:06.295Z","lastSeenAt":"2026-05-18T19:14:26.045Z"}],"details":{"listingId":"df4565cb-721e-43aa-9ba5-4c5c7c95d7dd","quickStartSnippet":null,"exampleRequest":null,"exampleResponse":null,"schema":null,"openapiUrl":null,"agentsTxtUrl":null,"citations":[],"useCases":[],"bestFor":[],"notFor":[],"kindDetails":{"org":"kambleakash0","slug":"english-humanizer","github":{"repo":"kambleakash0/agent-skills","stars":6,"topics":["agent-skills","ai-agents","ai-assistant","antigravity","automation","claude-code","code-review","codex","cusror","developer-tools","gemini-cli","llm-tools","mcp","prompt-engineering"],"license":"mit","html_url":"https://github.com/kambleakash0/agent-skills","pushed_at":"2026-05-02T07:18:13Z","description":"A curated collection of custom AgentSkills and token-saving MCP servers for extending the capabilities of AI coding assistants like Claude Code.","skill_md_sha":"ba16ccb2a1e9c33c2033059c0f5479f5fde98692","skill_md_path":"skills/english-humanizer/SKILL.md","default_branch":"main","skill_tree_url":"https://github.com/kambleakash0/agent-skills/tree/main/skills/english-humanizer"},"layout":"multi","source":"github","category":"agent-skills","frontmatter":{"name":"english-humanizer","license":"MIT","description":"Detects and removes AI-generated writing patterns from English text. Rewrites content to sound natural, authentic, and genuinely human."},"skills_sh_url":"https://skills.sh/kambleakash0/agent-skills/english-humanizer"},"updatedAt":"2026-05-18T19:14:26.045Z"}}