{"id":"5447a738-188f-417f-9db2-36abf9eb405a","shortId":"Gwtqxs","kind":"skill","title":"parallel-power","tagline":"Shows you how to use Cowork's parallel workers to run multiple tasks at the same time instead of one after another — dramatically cutting the time on batch work. For processing large batches of files or research tasks, when running multiple independent analyses simultaneously, or","description":"# Parallel Power\n\n## Role\n\nYou are a Cowork performance coach. You show users how to think in parallel —\nrestructuring their requests so Cowork handles multiple workstreams at once\ninstead of waiting for each one to finish before starting the next.\n\n## Why This Skill Exists\n\nBy default, most users ask Cowork for things one at a time: \"Write a summary\nof file 1.\" Wait. \"Now do file 2.\" Wait. For 10 files, that means 10 wait\ncycles. Parallel workers eliminate this. When Cowork spawns multiple workers,\neach one runs independently at the same time. Ten summaries that would take\n30 minutes serially finish in 3–4 minutes in parallel.\n\nThis is one of Cowork's most powerful features. Almost no one uses it.\n\n## Instructions\n\n### Step 1 — Identify whether the user has a parallel-ready task\n\nA task is parallel-ready when it can be split into independent pieces that do\nnot depend on each other's results. Ask:\n\n\"What are you trying to get done? Walk me through the work.\"\n\nGood candidates for parallel processing:\n- Summarizing, analyzing, or reformatting multiple files\n- Researching several companies, topics, or competitors\n- Drafting variations of the same document (different audiences, tones)\n- Processing rows or entries in a list where each row is independent\n\nNot suitable for parallel processing:\n- Tasks where Step 2 requires the result of Step 1\n- Tasks that write to the same file simultaneously\n- Tasks that require real-time conversation with the user\n\nIf the task is not parallel-ready, explain why and suggest the staging\napproach from the context-manager skill instead.\n\n### Step 2 — Show the time comparison\n\nMake the benefit concrete. Before asking Cowork to run anything in parallel,\nestimate the serial time vs the parallel time:\n\n\"If Cowork processes each of your 8 files one at a time and each takes\nabout 3 minutes, that's 24 minutes total. With parallel workers, all 8 run\nat the same time — closer to 3–4 minutes. That's an 80% reduction.\"\n\nAdjust the numbers for the user's actual task. Even rough estimates land\nthe point.\n\n### Step 3 — Restructure their request for parallel execution\n\nRewrite their original request so it explicitly asks for parallel workers.\nThe key phrase is: \"Use parallel workers to handle each [item] simultaneously.\"\n\nTemplate:\n> \"I have [N] [items]. Use parallel workers to [task] for each one\n> simultaneously. Save each result to a separate file named [pattern].\n> When all workers are done, combine the results into a single summary\n> file called [output-name].md.\"\n\nExample (before): \"Summarize each of these 6 competitor websites.\"\nExample (after): \"I have 6 competitor websites listed below. Use parallel\nworkers to visit each site and write a 200-word summary of their positioning\nand pricing. Save each summary as competitor-[name].md. When all are done,\ncreate a file called competitive-overview.md that combines the six summaries\ninto a comparison table.\"\n\n### Step 4 — Run the parallel task together\n\nHave the user paste the restructured prompt into Cowork now. Watch what\nhappens. Point out:\n\n- The workers spinning up (Cowork will show multiple tasks running)\n- How long it takes vs how long the serial version would have taken\n- The output files being created\n\nIf any worker fails or produces poor output, note which one and explain\nhow to rerun just that piece without redoing the whole batch.\n\n### Step 5 — Teach the merging step\n\nParallel workers produce separate outputs. The final step is always a merge:\n\n\"Now tell Cowork: 'Read all the [output files] and combine them into a\nsingle [format] saved as [final-filename].md.' This is the consolidation\nstep — it runs serially, but it only happens once, so it's fast.\"\n\nHelp the user write their specific merge prompt.\n\n### Step 6 — Give them the parallel-ready checklist\n\nLeave them with a mental checklist they can apply to future tasks:\n\n1. Can this be split into N independent pieces? If yes, use parallel workers.\n2. Does each piece produce a file as its output? If yes, add a merge step.\n3. Are the pieces roughly the same size? If wildly different, split the\n   biggest piece further before running.\n4. Is N greater than 5? Parallel workers save the most time on 5+ items.\n\n## Quality Checks\n\nBefore finishing, verify:\n\n- [ ] The user's task was confirmed as parallel-ready (independent pieces)\n- [ ] A time comparison was given (serial vs parallel estimate)\n- [ ] The request was rewritten using the parallel worker template\n- [ ] The parallel task was run at least once during the session\n- [ ] The merge step was explained and a merge prompt was written\n- [ ] The user understands the checklist for identifying future parallel tasks\n- [ ] Output quality was reviewed — not just speed\n\n## Examples\n\n**Example 1 — Summarizing a folder of competitor research files:**\nUser types: \"process multiple files — I have 8 competitor reports and it's taking forever\"\nCowork estimates the time comparison (24 minutes serial vs 3-4 minutes parallel), rewrites the request using the parallel worker template (\"Use parallel workers to summarize each of the 8 files simultaneously, save each as competitor-[name]-summary.md, then combine into competitive-overview.md\"), and runs it live while the user watches multiple workers execute at once.\n\n**Example 2 — Drafting content for multiple audiences:**\nUser types: \"do several things at once — write this announcement for five different audiences\"\nCowork identifies this as a parallel-ready task (five independent drafts that don't depend on each other), restructures the prompt, runs all five simultaneously, and then runs a single merge step to collect them into one document with labeled sections for each audience.\n\n**Example 3 — Processing a spreadsheet of contacts:**\nUser types: \"batch task — I have 12 companies I need quick research summaries on\"\nCowork confirms the task is parallel-ready (each company is independent), estimates 36 minutes serially vs 4 minutes with parallel workers, writes the structured parallel prompt, and runs it — producing 12 individual files plus a combined comparison table as the final output.\n\n## Troubleshooting\n\n**Issue: One or two workers fail while the others complete successfully.**\nSolution: Don't rerun the entire batch. Identify which items failed from the output file list, then send a targeted follow-up: \"Process only these two items using the same parallel worker approach and save them with the same filename pattern as the others.\" Then run the merge step again with all files including the new ones.\n\n**Issue: The merged output doesn't reflect all the parallel outputs.**\nSolution: The merge prompt needs to explicitly list the file pattern to read — if files are named inconsistently, the merge step may miss some. Rewrite the merge instruction with an explicit glob pattern: \"Read all files matching the pattern competitor-*.md in the [folder] and combine them.\"\n\n**Issue: User tries parallel processing on a task that actually has sequential dependencies.**\nSolution: Step 1 of the skill is the safeguard — but if it's discovered mid-run, stop and reframe. Explain which steps depend on previous results, then split the task: run the dependent steps serially first to produce the shared input, then use parallel workers for the independent downstream steps that all share that same input.\n\n## Related Skills\n\nSee also: **workflow-builder** — many multi-step workflows contain a batch phase that can be parallelized; redesign those phases with parallel workers for a major speed gain.\nRelated: **dispatch-starter** — large parallel batch jobs can be triggered via Dispatch so they run on your desktop while you're away, with results waiting when you return.\nSee also: **skill-creator-guide** — if you run the same parallel batch pattern regularly, package it as a custom skill so you trigger it with a phrase instead of rewriting the prompt each time.","tags":["parallel","power","cowork","skills","library","eaiconsulting","agent-skills","ai-automation","ai-skills","anthropic-claude","claude","claude-code-plugin"],"capabilities":["skill","source-eaiconsulting","skill-parallel-power","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/parallel-power","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 (7,874 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":"2026-05-18T19:13:34.418Z","embedding":null,"createdAt":"2026-05-18T13:21:05.617Z","updatedAt":"2026-05-18T19:13:34.418Z","lastSeenAt":"2026-05-18T19:13:34.418Z","tsv":"'-4':860 '1':109,172,271,688,827,1178 '10':117,121 '12':984,1023 '2':114,265,313,702,906 '200':496 '24':358,855 '3':151,354,373,397,718,859,972 '30':146 '36':1005 '4':152,374,530,736,1009 '5':604,741,749 '6':474,481,668 '8':344,365,842,879 '80':379 'actual':388,1172 'add':714 'adjust':381 'almost':165 'also':1236,1294 'alway':618 'analys':46 'analyz':225 'announc':921 'anoth':25 'anyth':327 'appli':684 'approach':304,1080 'ask':96,206,323,411 'audienc':243,911,925,970 'away':1286 'batch':31,36,602,980,1053,1247,1270,1305 'benefit':320 'biggest':731 'builder':1239 'call':463,518 'candid':220 'check':752 'checklist':675,681,812 'closer':371 'coach':57 'collect':960 'combin':455,521,630,889,1028,1161 'compani':232,985,1001 'comparison':317,527,770,854,1029 'competitive-overview.md':519,891 'competitor':235,475,482,508,832,843,885,1155 'complet':1045 'concret':321 'confirm':761,993 'consolid':645 'contact':977 'contain':1245 'content':908 'context':308 'context-manag':307 'convers':286 'cowork':9,55,70,97,129,160,324,339,544,555,623,850,926,992 'creat':515,578 'creator':1297 'custom':1312 'cut':27 'cycl':123 'default':93 'depend':200,941,1175,1199,1209 'desktop':1282 'differ':242,728,924 'discov':1189 'dispatch':1266,1276 'dispatch-start':1265 'document':241,964 'doesn':1109 'done':213,454,514 'downstream':1225 'draft':236,907,937 'dramat':26 'elimin':126 'entir':1052 'entri':248 'estim':330,392,776,851,1004 'even':390 'exampl':468,477,825,826,905,971 'execut':403,902 'exist':91 'explain':298,591,801,1196 'explicit':410,1122,1146 'fail':582,1041,1057 'fast':658 'featur':164 'file':38,108,113,118,229,278,345,447,462,517,576,628,708,834,839,880,1025,1061,1100,1125,1130,1151 'filenam':640,1087 'final':615,639,1033 'final-filenam':638 'finish':83,149,754 'first':1212 'five':923,935,950 'folder':830,1159 'follow':1068 'follow-up':1067 'forev':849 'format':635 'futur':686,815 'gain':1263 'get':212 'give':669 'given':772 'glob':1147 'good':219 'greater':739 'guid':1298 'handl':71,423 'happen':548,653 'help':659 'identifi':173,814,927,1054 'includ':1101 'inconsist':1133 'independ':45,136,195,256,695,766,936,1003,1224 'individu':1024 'input':1217,1232 'instead':21,76,311,1321 'instruct':170,1143 'issu':1036,1105,1163 'item':425,431,750,1056,1074 'job':1271 'key':416 'label':966 'land':393 'larg':35,1268 'least':792 'leav':676 'list':251,484,1062,1123 'live':895 'long':562,567 'major':1261 'make':318 'manag':309 'mani':1240 'match':1152 'may':1137 'md':467,510,641,1156 'mean':120 'mental':680 'merg':607,620,665,716,798,804,957,1095,1107,1118,1135,1142 'mid':1191 'mid-run':1190 'minut':147,153,355,359,375,856,861,1006,1010 'miss':1138 'multi':1242 'multi-step':1241 'multipl':15,44,72,131,228,558,838,900,910 'n':430,694,738 'name':448,466,509,886,1132 'need':987,1120 'new':1103 'next':87 'note':587 'number':383 'one':23,81,100,134,158,167,346,439,589,963,1037,1104 'origin':406 'other':1044,1091 'output':465,575,586,613,627,711,818,1034,1060,1108,1115 'output-nam':464 'packag':1308 'parallel':2,11,49,65,124,155,180,187,222,260,296,329,336,362,402,413,420,433,487,533,609,673,700,742,764,775,783,787,816,862,868,872,932,998,1012,1017,1078,1114,1166,1220,1252,1257,1269,1304 'parallel-pow':1 'parallel-readi':179,186,295,672,763,931,997 'past':539 'pattern':449,1088,1126,1148,1154,1306 'perform':56 'phase':1248,1255 'phrase':417,1320 'piec':196,597,696,705,721,732,767 'plus':1026 'point':395,549 'poor':585 'posit':501 'power':3,50,163 'previous':1201 'price':503 'process':34,223,245,261,340,837,973,1070,1167 'produc':584,611,706,1022,1214 'prompt':542,666,805,947,1018,1119,1325 'qualiti':751,819 'quick':988 're':1285 'read':624,1128,1149 'readi':181,188,297,674,765,933,999 'real':284 'real-tim':283 'redesign':1253 'redo':599 'reduct':380 'reflect':1111 'reformat':227 'refram':1195 'regular':1307 'relat':1233,1264 'report':844 'request':68,400,407,778,865 'requir':266,282 'rerun':594,1050 'research':40,230,833,989 'restructur':66,398,541,945 'result':205,268,443,457,1202,1288 'return':1292 'review':821 'rewrit':404,863,1140,1323 'rewritten':780 'role':51 'rough':391,722 'row':246,254 'run':14,43,135,326,366,531,560,648,735,790,893,948,954,1020,1093,1192,1207,1279,1301 'safeguard':1184 'save':441,504,636,744,882,1082 'section':967 'see':1235,1293 'send':1064 'separ':446,612 'sequenti':1174 'serial':148,332,569,649,773,857,1007,1211 'session':796 'sever':231,915 'share':1216,1229 'show':4,59,314,557 'simultan':47,279,426,440,881,951 'singl':460,634,956 'site':492 'six':523 'size':725 'skill':90,310,1181,1234,1296,1313 'skill-creator-guid':1295 'skill-parallel-power' 'solut':1047,1116,1176 'source-eaiconsulting' 'spawn':130 'specif':664 'speed':824,1262 'spin':553 'split':193,692,729,1204 'spreadsheet':975 'stage':303 'start':85 'starter':1267 'step':171,264,270,312,396,529,603,608,616,646,667,717,799,958,1096,1136,1177,1198,1210,1226,1243 'stop':1193 'structur':1016 'success':1046 'suggest':301 'suitabl':258 'summar':224,470,828,875 'summari':106,142,461,498,506,524,990 'summary.md':887 'tabl':528,1030 'take':145,352,564,848 'taken':573 'target':1066 'task':16,41,182,184,262,272,280,292,389,436,534,559,687,759,788,817,934,981,995,1170,1206 'teach':605 'tell':622 'templat':427,785,870 'ten':141 'thing':99,916 'think':63 'time':20,29,103,140,285,316,333,337,349,370,747,769,853,1327 'togeth':535 'tone':244 'topic':233 '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' 'total':360 'tri':210,1165 'trigger':1274,1316 'troubleshoot':1035 'two':1039,1073 'type':836,913,979 'understand':810 'use':8,168,419,432,486,699,781,866,871,1075,1219 'user':60,95,176,289,386,538,661,757,809,835,898,912,978,1164 'variat':237 'verifi':755 'version':570 'via':1275 'visit':490 'vs':334,565,774,858,1008 'wait':78,110,115,122,1289 'walk':214 'watch':546,899 'websit':476,483 'whether':174 'whole':601 'wild':727 'without':598 'word':497 'work':32,218 'worker':12,125,132,363,414,421,434,452,488,552,581,610,701,743,784,869,873,901,1013,1040,1079,1221,1258 'workflow':1238,1244 'workflow-build':1237 'workstream':73 'would':144,571 'write':104,274,494,662,919,1014 'written':807 'yes':698,713","prices":[{"id":"605496b4-9986-4ffa-a6e6-2a14a7e4a58c","listingId":"5447a738-188f-417f-9db2-36abf9eb405a","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":"2026-05-18T13:21:05.617Z"}],"sources":[{"listingId":"5447a738-188f-417f-9db2-36abf9eb405a","source":"github","sourceId":"EAIconsulting/cowork-skills-library/parallel-power","sourceUrl":"https://github.com/EAIconsulting/cowork-skills-library/tree/main/skills/parallel-power","isPrimary":false,"firstSeenAt":"2026-05-18T13:21:05.617Z","lastSeenAt":"2026-05-18T19:13:34.418Z"}],"details":{"listingId":"5447a738-188f-417f-9db2-36abf9eb405a","quickStartSnippet":null,"exampleRequest":null,"exampleResponse":null,"schema":null,"openapiUrl":null,"agentsTxtUrl":null,"citations":[],"useCases":[],"bestFor":[],"notFor":[],"kindDetails":{"org":"EAIconsulting","slug":"parallel-power","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":"e2e99ce95cdc62a2d20453555166dc51d8278a59","skill_md_path":"skills/parallel-power/SKILL.md","default_branch":"main","skill_tree_url":"https://github.com/EAIconsulting/cowork-skills-library/tree/main/skills/parallel-power"},"layout":"multi","source":"github","category":"cowork-skills-library","frontmatter":{"name":"parallel-power","description":"Shows you how to use Cowork's parallel workers to run multiple tasks at the same time instead of one after another — dramatically cutting the time on batch work. For processing large batches of files or research tasks, when running multiple independent analyses simultaneously, or during a performance optimization session. Use when the user says \"speed up\", \"parallel workers\", \"batch task\", \"process multiple files\", \"why is cowork so slow\", \"faster\", \"do several things at once\", or \"can cowork multitask\"."},"skills_sh_url":"https://skills.sh/EAIconsulting/cowork-skills-library/parallel-power"},"updatedAt":"2026-05-18T19:13:34.418Z"}}