{"id":"af292bee-e727-4f7e-98a1-108118ab65a3","shortId":"PCRbp5","kind":"skill","title":"teach-your-voice","tagline":"'Builds a personal voice profile so Cowork writes the way you write —","description":"# Teach Claude Your Voice\n\n**Your Role:** You are a writing coach and communication specialist who helps\nprofessionals sound like themselves — even when an AI is doing the drafting. You\nbelieve that the best AI-assisted writing is indistinguishable from the person's\nown voice. You are perceptive, warm, and genuinely interested in how each person\ncommunicates.\n\n**Goal:** Build a voice profile that captures how the user thinks, writes, and\ncommunicates — then save it where Cowork can use it automatically on every future\nwriting task.\n\n## Why This Skill Exists\n\nWhen Cowork writes something for you — an email, a proposal, a Slack message —\nthe default output sounds like a polished but soulless assistant. Complete sentences.\nBalanced paragraphs. Correct but characterless.\n\nThe problem: Cowork doesn't know how you write. It doesn't know you use short\nsentences under pressure. That you start emails with the conclusion, not the setup.\nThat you never say \"please don't hesitate to reach out.\" That you make dry jokes\nin internal messages but stay measured with clients.\n\nThis skill fixes that. It interviews you about your communication style, optionally\nanalyzes samples of your actual writing, and creates a voice profile. Once saved,\nCowork will write in your voice — not its default voice — for every email, message,\ndocument, and summary it produces.\n\n## Instructions\n\n### Step 1: Open the Conversation\n\nSay:\n\n\"Let's teach Cowork to write the way you write. This takes about 10 minutes and the\nresult is that future drafts will sound like you — not like a well-behaved AI.\n\nI'll ask you some questions about how you communicate, and if you want, you can\npaste in some writing samples so I can analyze your actual patterns. Let's start.\"\n\n### Step 2: Run the Voice Interview\n\nAsk these questions one at a time, in a conversational way. Listen carefully —\nthe goal is to capture real patterns, not ideal self-descriptions.\n\n**Question 1 — The gut check:**\n\"Read this sentence: 'Please find attached the report summarizing our Q3 performance,\nwhich I believe you will find informative.' Would you ever write something like that?\nWhat would you actually write instead?\"\n\n(This question reveals formality level instantly. Their rewrite tells you more\nthan any self-description.)\n\n**Question 2 — Length and density:**\n\"When you write a work email, how long is it usually? Do you tend to write in full\nparagraphs, bullets, or a mix? What does a typical internal message look like?\"\n\n**Question 3 — How you open:**\n\"How do you usually start an email to someone you know well at work? To a client\nor senior stakeholder you want to impress? Any phrases you always use — or ones\nyou'd never be caught dead writing?\"\n\n**Question 4 — Personality and tone:**\n\"Does your writing personality vary a lot depending on the audience? How would you\ndescribe the difference between how you write to your team versus to a client versus\nto your boss?\"\n\n**Question 5 — What you hate:**\n\"What are your biggest pet peeves about AI-generated writing? (Think: phrases that\nmake you immediately edit, things that feel fake, formats that annoy you.)\"\n\n**Question 6 — Your actual words:**\n\"Are there any specific words, phrases, or expressions you use a lot that feel\ndistinctly 'you'? And any you'd never use?\"\n\n### Step 3: Offer to Analyze Writing Samples (Optional)\n\nSay:\n\n\"One optional step that makes the voice profile much sharper: you can paste in\n2-3 examples of writing you're proud of — emails, messages, short documents,\nanything where you think 'yes, that sounds like me.' I'll analyze the patterns\nand add them to your profile.\n\nDo you want to do that, or is the interview enough for now?\"\n\nIf they share samples, analyze for:\n- Sentence length distribution (short vs. long, mix)\n- Paragraph length and structure\n- Use of bullet points vs. prose\n- Vocabulary level and formality markers\n- Common phrases and openers\n- Punctuation habits (em dashes, ellipses, etc.)\n- What they never say (negative space is equally important)\n- Emotional register (warm? measured? direct? playful?)\n\nSummarize what you find before building the profile.\n\n### Step 4: Build the Voice Profile\n\nCreate a file called `my-writing-voice.md` in their references folder\n(or their home directory if no references folder exists yet).\n\nStructure it as:\n\n```\n# My Writing Voice\n\n## The One-Sentence Summary\n[A single sentence that captures their core communication style. Should feel like\na description a close colleague would give. Example: \"Direct and warm — gets to\nthe point fast but always sounds like a real person, not a memo.\"]\n\n## Tone and Register\n[2-4 sentences on their overall tone. Formal/informal spectrum, warmth level,\nwhether they use humor and how, whether they vary by audience.]\n\n## How I Structure Writing\n[Bullet list: how they open, how they organize information, how long things are,\nbullets vs. prose preference, how they close.]\n\n## Words and Phrases I Use\n[Honest list of recurring vocabulary, openers, and expressions that feel like them.\nPull from their samples or answers.]\n\n## Words and Phrases I Avoid\n[List of phrases they hate or would never write. Include AI clichés they mentioned.\nThis section is just as important as the previous one.]\n\n## By Audience\n[If they communicate very differently with different groups, note the key differences\nhere: internal team vs. clients vs. leadership vs. external stakeholders.]\n\n## Quick Test\n[2-3 sentences that could only be from them. This is the \"does it pass the voice\ntest\" calibration tool — if Cowork writes something that sounds nothing like these\nsentences, it needs to adjust.]\n```\n\n### Step 5: Activate the Voice Profile\n\nSay:\n\n\"Your voice profile is saved at `my-writing-voice.md`. There are two ways to use it:\n\n**Option 1 — Always on (recommended):**\nOpen Cowork's Settings, find the Preferences section, and add this line at the end:\n'When writing anything for me — emails, messages, summaries, documents — read my\nvoice profile at [path to my-writing-voice.md] and match my style.'\n\n**Option 2 — Task-by-task:**\nWhen you want Cowork to write in your voice, start with: 'Write this in my voice\n— check my-writing-voice.md first.'\"\n\n### Step 6: Run a Live Test\n\nPick a quick writing task based on what they told you — something concrete. For\nexample, if they mentioned client emails were a priority, offer to draft one right now\nusing the voice profile.\n\nSay: \"Let's test it. Give me a writing task you'd normally do yourself — an email\nto reply to, a message to draft, anything. I'll write it using your voice profile\nand you can tell me how close it is.\"\n\nAfter drafting, ask: \"On a scale from 1-5, how much does that sound like you?\nWhat would you change?\" Adjust the profile based on their feedback.\n\n### Step 7: Deliver the Summary\n\nSay:\n\n\"Your voice profile is complete and active. Here's what's saved in\n`my-writing-voice.md`:\n\n- Your tone and communication style\n- How you structure writing for different audiences\n- Phrases you use — and ones you'd never be caught using\n- A quick test sample to calibrate against\n\nEvery writing task you give Cowork from now on can reference this profile. The\nmore writing tasks you delegate, the more consistent the voice becomes. If it drifts,\njust say 'check my voice profile and adjust' and it will recalibrate.\"\n\n## Quality Checks\n\nBefore finishing:\n- [ ] Voice profile is specific, not generic (\"direct and uses bullet points\" is\n  not enough — it should have actual phrases and examples)\n- [ ] \"Words I avoid\" section is populated — this is the most common gap\n- [ ] If samples were provided, specific patterns from those samples are reflected\n- [ ] The Quick Test sentences in the profile could genuinely only be from this person\n- [ ] Live test was offered and feedback was incorporated\n- [ ] User was given both activation options (always-on and task-by-task)\n- [ ] Profile avoids describing an idealized self — it captures how they actually write\n- [ ] No technical jargon used (no \"system prompt\", \"context injection\", \"token\")\n\n## Examples\n\n**Example 1:** \"Cowork writes in complete, formal sentences but I'm very direct — fix that\"\n→ Skill runs the 6-question voice interview, captures the user's preference for short direct sentences, identifies specific phrases they would never write, and produces a voice profile with a Quick Test calibration sample.\n\n**Example 2:** \"I want to paste in some emails I've written — can you figure out my style?\"\n→ Skill analyzes the samples for sentence length, paragraph structure, bullet vs prose preference, vocabulary level, and punctuation habits — then builds a voice profile grounded in their actual writing patterns.\n\n**Example 3:** \"I write differently for my team than for clients — can Cowork handle both?\"\n→ Skill captures both modes in the \"By Audience\" section of the voice profile, and explains how to invoke each one: \"Write this in my internal team voice\" vs \"Write this in my client-facing voice.\"\n\n## Troubleshooting\n\n**Skill didn't activate when I described my task?**\nTry invoking directly with `/teach-your-voice` to start the voice interview.\n\n**The voice profile didn't stick — output still sounds generic?**\nMake sure you activated the profile using Option 1 (always-on) by adding the reference line to Cowork's Preferences settings. Task-by-task activation (Option 2) requires you to reference the profile explicitly at the start of each writing task.\n\n**The live test sounded close but not quite right?**\nShare specific feedback during the live test (\"that phrase sounds too formal\" or \"I'd never start a sentence that way\"). The profile is refined based on your feedback — one round of calibration usually closes the gap.\n\n## Related Skills\n\nSee also:\n- `/cowork-setup-wizard` — Set up your full Cowork preferences before or alongside your voice profile\n- `/delegation-coach` — Learn how to give Cowork clear instructions so voice-matched writing tasks deliver well\n- `/context-manager` — Save your voice profile where Cowork can reference it across multi-session projects","tags":["teach","your","voice","cowork","skills","library","eaiconsulting","agent-skills","ai-automation","ai-skills","anthropic-claude","claude"],"capabilities":["skill","source-eaiconsulting","skill-teach-your-voice","topic-agent-skills","topic-ai-automation","topic-ai-skills","topic-anthropic-claude","topic-claude","topic-claude-code-plugin","topic-claude-code-skills","topic-claude-cowork","topic-claude-desktop","topic-claude-skills","topic-cowork","topic-knowledge-work"],"categories":["cowork-skills-library"],"synonyms":[],"warnings":[],"endpointUrl":"https://skills.sh/EAIconsulting/cowork-skills-library/teach-your-voice","protocol":"skill","transport":"skills-sh","auth":{"type":"none","details":{"cli":"npx skills add EAIconsulting/cowork-skills-library","source_repo":"https://github.com/EAIconsulting/cowork-skills-library","install_from":"skills.sh"}},"qualityScore":"0.453","qualityRationale":"deterministic score 0.45 from registry signals: · indexed on github topic:agent-skills · 7 github stars · SKILL.md body (10,042 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":"2026-05-18T19:13:35.062Z","embedding":null,"createdAt":"2026-05-18T13:21:06.360Z","updatedAt":"2026-05-18T19:13:35.062Z","lastSeenAt":"2026-05-18T19:13:35.062Z","tsv":"'-3':586,894 '-4':770 '-5':1103 '/context-manager':1618 '/cowork-setup-wizard':1589 '/delegation-coach':1602 '/teach-your-voice':1480 '1':235,336,949,1102,1320,1504 '10':253 '2':305,389,585,769,893,990,1369,1524 '3':425,563,1416 '4':468,693 '5':505,928 '6':536,1015,1337 '7':1123 'across':1628 'activ':929,1134,1286,1470,1499,1522 'actual':205,299,369,538,1233,1306,1412 'ad':1509 'add':613,962 'adjust':926,1115,1207 'ai':40,51,272,517,853 'ai-assist':50 'ai-gener':516 'alongsid':1598 'also':1588 'alway':456,757,950,1289,1506 'always-on':1288,1505 'analyz':201,297,566,609,635,1387 'annoy':533 'answer':837 'anyth':598,970,1077 'ask':275,310,1097 'assist':52,128 'attach':345 'audienc':482,790,868,1153,1437 'automat':96 'avoid':842,1239,1297 'balanc':131 'base':1025,1118,1573 'becom':1196 'behav':271 'believ':46,354 'best':49 'biggest':512 'boss':503 'build':5,75,689,694,1405 'bullet':412,650,795,808,1225,1395 'calibr':911,1170,1366,1580 'call':701 'captur':80,327,732,1303,1341,1431 'care':322 'caught':464,1163 'chang':1114 'characterless':135 'check':339,1011,1202,1213 'claud':18 'clear':1608 'cliché':854 'client':188,445,499,885,1038,1425,1463 'client-fac':1462 'close':743,814,1092,1543,1582 'coach':27 'colleagu':744 'common':659,1247 'communic':29,73,87,198,282,735,871,1145 'complet':129,1132,1324 'conclus':161 'concret':1032 'consist':1193 'context':1315 'convers':238,319 'core':734 'correct':133 'could':897,1267 'cowork':11,92,107,138,214,243,914,954,998,1177,1321,1427,1514,1594,1607,1624 'creat':208,698 'd':461,559,1064,1160,1562 'dash':666 'dead':465 'default':120,222 'deleg':1190 'deliv':1124,1616 'densiti':392 'depend':479 'describ':486,1298,1473 'descript':334,387,741 'didn':1468,1489 'differ':488,873,875,880,1152,1419 'direct':682,748,1222,1331,1348,1478 'directori':710 'distinct':554 'distribut':639 'document':228,597,976 'doesn':139,146 'draft':44,261,1045,1076,1096 'dri':179 'drift':1199 'edit':526 'ellips':667 'em':665 'email':113,158,226,398,435,594,973,1039,1069,1376 'emot':678 'end':967 'enough':628,1229 'equal':676 'etc':668 'even':37 'ever':361 'everi':98,225,1172 'exampl':587,747,1034,1236,1318,1319,1368,1415 'exist':105,715 'explain':1444 'explicit':1531 'express':547,827 'extern':889 'face':1464 'fake':530 'fast':755 'feedback':1121,1279,1550,1576 'feel':529,553,738,829 'figur':1382 'file':700 'find':344,357,687,957 'finish':1215 'first':1013 'fix':191,1332 'folder':706,714 'formal':375,657,1325,1559 'formal/informal':776 'format':531 'full':410,1593 'futur':99,260 'gap':1248,1584 'generat':518 'generic':1221,1495 'genuin':67,1268 'get':751 'give':746,1058,1176,1606 'given':1284 'goal':74,324 'ground':1409 'group':876 'gut':338 'habit':664,1403 'handl':1428 'hate':508,847 'help':32 'hesit':172 'home':709 'honest':820 'humor':783 'ideal':331,1300 'identifi':1350 'immedi':525 'import':677,862 'impress':452 'includ':852 'incorpor':1281 'indistinguish':55 'inform':358,803 'inject':1316 'instant':377 'instead':371 'instruct':233,1609 'interest':68 'intern':182,420,882,1454 'interview':194,309,627,1340,1485 'invok':1447,1477 'jargon':1310 'joke':180 'key':879 'know':141,148,439 'leadership':887 'learn':1603 'length':390,638,645,1392 'let':240,301,1054 'level':376,655,779,1400 'like':35,123,264,267,364,423,605,739,759,830,920,1109 'line':964,1512 'list':796,821,843 'listen':321 'live':1018,1274,1540,1553 'll':274,608,1079 'long':400,642,805 'look':422 'lot':478,551 'm':1329 'make':178,523,575,1496 'marker':658 'match':986,1613 'measur':186,681 'memo':765 'mention':856,1037 'messag':118,183,227,421,595,974,1074 'minut':254 'mix':415,643 'mode':1433 'much':579,1105 'multi':1630 'multi-sess':1629 'my-writing-voice.md':702,940,984,1012,1141 'need':924 'negat':673 'never':167,462,560,671,850,1161,1355,1563 'normal':1065 'note':877 'noth':919 'offer':564,1043,1277 'one':313,459,571,725,866,1046,1158,1449,1577 'one-sent':724 'open':236,428,662,799,825,953 'option':200,569,572,948,989,1287,1503,1523 'organ':802 'output':121,1492 'overal':774 'paragraph':132,411,644,1393 'pass':907 'past':289,583,1373 'path':982 'pattern':300,329,611,1254,1414 'peev':514 'percept':64 'perform':351 'person':7,58,72,469,475,762,1273 'pet':513 'phrase':454,521,545,660,817,840,845,1154,1234,1352,1556 'pick':1020 'play':683 'pleas':169,343 'point':651,754,1226 'polish':125 'popul':1242 'prefer':811,959,1345,1398,1516,1595 'pressur':154 'previous':865 'prioriti':1042 'problem':137 'produc':232,1358 'profession':33 'profil':9,78,211,578,617,691,697,932,936,980,1052,1085,1117,1130,1184,1205,1217,1266,1296,1361,1408,1442,1488,1501,1530,1570,1601,1622 'project':1632 'prompt':1314 'propos':115 'prose':653,810,1397 'proud':592 'provid':1252 'pull':832 'punctuat':663,1402 'q3':350 'qualiti':1212 'question':278,312,335,373,388,424,467,504,535,1338 'quick':891,1022,1166,1261,1364 'quit':1546 're':591 'reach':174 'read':340,977 'real':328,761 'recalibr':1211 'recommend':952 'recur':823 'refer':705,713,1182,1511,1528,1626 'refin':1572 'reflect':1259 'regist':679,768 'relat':1585 'repli':1071 'report':347 'requir':1525 'result':257 'reveal':374 'rewrit':379 'right':1047,1547 'role':22 'round':1578 'run':306,1016,1335 'sampl':202,293,568,634,835,1168,1250,1257,1367,1389 'save':89,213,938,1139,1619 'say':168,239,570,672,933,1053,1127,1201 'scale':1100 'section':858,960,1240,1438 'see':1587 'self':333,386,1301 'self-descript':332,385 'senior':447 'sentenc':130,152,342,637,726,730,771,895,922,1263,1326,1349,1391,1566 'session':1631 'set':956,1517,1590 'setup':164 'share':633,1548 'sharper':580 'short':151,596,640,1347 'singl':729 'skill':104,190,1334,1386,1430,1467,1586 'skill-teach-your-voice' 'slack':117 'someon':437 'someth':109,363,916,1031 'soulless':127 'sound':34,122,263,604,758,918,1108,1494,1542,1557 'source-eaiconsulting' 'space':674 'specialist':30 'specif':543,1219,1253,1351,1549 'spectrum':777 'stakehold':448,890 'start':157,303,433,1004,1482,1534,1564 'stay':185 'step':234,304,562,573,692,927,1014,1122 'stick':1491 'still':1493 'structur':647,717,793,1149,1394 'style':199,736,988,1146,1385 'summar':348,684 'summari':230,727,975,1126 'sure':1497 'system':1313 'take':251 'task':101,992,994,1024,1062,1174,1188,1293,1295,1475,1519,1521,1538,1615 'task-by-task':991,1292,1518 'teach':2,17,242 'teach-your-voic':1 'team':495,883,1422,1455 'technic':1309 'tell':380,1089 'tend':406 'test':892,910,1019,1056,1167,1262,1275,1365,1541,1554 'thing':527,806 'think':84,520,601 'time':316 'token':1317 'told':1029 'tone':471,766,775,1143 'tool':912 'topic-agent-skills' 'topic-ai-automation' 'topic-ai-skills' 'topic-anthropic-claude' 'topic-claude' 'topic-claude-code-plugin' 'topic-claude-code-skills' 'topic-claude-cowork' 'topic-claude-desktop' 'topic-claude-skills' 'topic-cowork' 'topic-knowledge-work' 'tri':1476 'troubleshoot':1466 'two':943 'typic':419 'use':94,150,457,549,561,648,782,819,946,1049,1082,1156,1164,1224,1311,1502 'user':83,1282,1343 'usual':403,432,1581 'vari':476,788 've':1378 'versus':496,500 'vocabulari':654,824,1399 'voic':4,8,20,61,77,210,219,223,308,577,696,722,909,931,935,979,1003,1010,1051,1084,1129,1195,1204,1216,1339,1360,1407,1441,1456,1465,1484,1487,1600,1612,1621 'voice-match':1611 'vs':641,652,809,884,886,888,1396,1457 'want':286,450,620,997,1371 'warm':65,680,750 'warmth':778 'way':14,247,320,944,1568 'well':270,440,1617 'well-behav':269 'whether':780,786 'word':539,544,815,838,1237 'work':397,442 'would':359,367,484,745,849,1112,1354 'write':12,16,26,53,85,100,108,144,206,216,245,249,292,362,370,395,408,466,474,492,519,567,589,721,794,851,915,969,1000,1006,1023,1061,1080,1150,1173,1187,1307,1322,1356,1413,1418,1450,1458,1537,1614 'written':1379 'yes':602 'yet':716","prices":[{"id":"c45d9d37-2ffc-4ca4-8a55-52d97e24c31d","listingId":"af292bee-e727-4f7e-98a1-108118ab65a3","amountUsd":"0","unit":"free","nativeCurrency":null,"nativeAmount":null,"chain":null,"payTo":null,"paymentMethod":"skill-free","isPrimary":true,"details":{"org":"EAIconsulting","category":"cowork-skills-library","install_from":"skills.sh"},"createdAt":"2026-05-18T13:21:06.360Z"}],"sources":[{"listingId":"af292bee-e727-4f7e-98a1-108118ab65a3","source":"github","sourceId":"EAIconsulting/cowork-skills-library/teach-your-voice","sourceUrl":"https://github.com/EAIconsulting/cowork-skills-library/tree/main/skills/teach-your-voice","isPrimary":false,"firstSeenAt":"2026-05-18T13:21:06.360Z","lastSeenAt":"2026-05-18T19:13:35.062Z"}],"details":{"listingId":"af292bee-e727-4f7e-98a1-108118ab65a3","quickStartSnippet":null,"exampleRequest":null,"exampleResponse":null,"schema":null,"openapiUrl":null,"agentsTxtUrl":null,"citations":[],"useCases":[],"bestFor":[],"notFor":[],"kindDetails":{"org":"EAIconsulting","slug":"teach-your-voice","github":{"repo":"EAIconsulting/cowork-skills-library","stars":7,"topics":["agent-skills","ai-automation","ai-skills","anthropic-claude","claude","claude-code-plugin","claude-code-skills","claude-cowork","claude-desktop","claude-skills","cowork","knowledge-work","no-code","productivity","skill-library"],"license":null,"html_url":"https://github.com/EAIconsulting/cowork-skills-library","pushed_at":"2026-04-09T01:46:19Z","description":"21 Claude Cowork skills: beginner to power user in 18 minutes. Free forever. By Everyday AI.","skill_md_sha":"53991de0cd90d693751df3fb031e8fdddd43248e","skill_md_path":"skills/teach-your-voice/SKILL.md","default_branch":"main","skill_tree_url":"https://github.com/EAIconsulting/cowork-skills-library/tree/main/skills/teach-your-voice"},"layout":"multi","source":"github","category":"cowork-skills-library","frontmatter":{"name":"teach-your-voice","description":"'Builds a personal voice profile so Cowork writes the way you write —"},"skills_sh_url":"https://skills.sh/EAIconsulting/cowork-skills-library/teach-your-voice"},"updatedAt":"2026-05-18T19:13:35.062Z"}}