[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"skill:01M71JWV3TE7CFJPSH8KPC5HR1":3,"skill-same:01M71JWV3TE7CFJPSH8KPC5HR1":120},{"skillNo":4,"name":5,"nameZh":6,"description":7,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":16,"categories":17,"keywords":23,"publishedAt":44,"linkOnly":45,"markdownContent":46,"markdownBody":47,"frontMatter":48,"refName":50,"commitSha":51,"skillDirectory":52,"forksCount":53,"repositoryHtmlUrl":54,"repositoryHomepageUrl":55,"ownerHtmlUrl":56,"ownerBlogUrl":57,"ownerDescription":58,"updatedAt":59,"install":60,"resources":73},"01M71JWV3TE7CFJPSH8KPC5HR1","maker-mdap",null,"Knowledge base from \"Solving a Million-Step LLM Task with Zero Errors\" (Meyerson, Paolo, Dailey, Shahrzad, Francon, Hayes, Qiu, Hodjat, Miikkulainen — arXiv:2511.09030). Use when designing long-horizon, multi-step agentic LLM pipelines that need very high reliability, applying Maximal Agentic Decomposition (MAD), first-to-ahead-by-k voting, or red-flagging error correction, estimating cost\u002Freliability scaling laws for multi-agent systems, or studying Massively Decomposed Agentic Processes (MDAPs).","coco-research\u002Fcoco","coco-research","https:\u002F\u002Favatars.githubusercontent.com\u002Fu\u002F68911604?v=4","01M71JRE9C4NM4NSZM4JD0BJTB","Coco Research","01M71JREA0ZMR0PMH3CKA99SN1",631,"Apache-2.0","https:\u002F\u002Fgithub.com\u002Fcoco-research\u002Fcoco\u002Fblob\u002Fmain\u002Fskills\u002Fmaker-mdap\u002FSKILL.md",[18],{"categoryNo":19,"categoryName":20,"taxonomyType":21,"taxonomyTypeDesc":22},"01JBSEEDCAT000000000000023","Skill与智能体开发","SCENARIO","使用场景",[24,25,26,27,28,29,30,31,32,33,34,35,36,37,38,39,40,41,42,43],"maker mdap","maker","scaling","when","horizon","for","errors","error correction","error","long","decomposition","agentic","knowledge base","arxiv","reliability","use","voting","multi-agent","knowledge","ZeRO","2026-11-10 10:12:30",false,"---\nname: maker-mdap\ndescription: \"Knowledge base from \\\"Solving a Million-Step LLM Task with Zero Errors\\\" (Meyerson, Paolo, Dailey, Shahrzad, Francon, Hayes, Qiu, Hodjat, Miikkulainen — arXiv:2511.09030). Use when designing long-horizon, multi-step agentic LLM pipelines that need very high reliability, applying Maximal Agentic Decomposition (MAD), first-to-ahead-by-k voting, or red-flagging error correction, estimating cost\u002Freliability scaling laws for multi-agent systems, or studying Massively Decomposed Agentic Processes (MDAPs).\"\ndomain: engineering\n---\n\n\u003C!-- argument-hint: [topic, framework name, or chapter number] -->\n\n# Solving a Million-Step LLM Task with Zero Errors\n**Authors**: Elliot Meyerson, Giuseppe Paolo, Roberto Dailey, Hormoz Shahrzad, Olivier Francon,\nConor F. Hayes, Xin Qiu, Babak Hodjat, Risto Miikkulainen (Cognizant AI Lab \u002F UT Austin)\n**Source**: [arXiv:2511.09030](https:\u002F\u002Farxiv.org\u002Fabs\u002F2511.09030) (Nov 2025) ([Notice](NOTICE.md)) | **Pages**: ~29 | **Chapters**: 7 | **Generated**: 2026-08-26\n\n## How to Use This Skill\n\n- **Without arguments** — load core frameworks below for reference\n- **With a topic** — ask about `voting`, `red-flagging`, `cost scaling`, `decomposition`, or\n  another indexed topic; I find and read the relevant chapter\n- **With a chapter** — ask for `ch03`; I load that specific chapter\n- **Browse** — ask \"what chapters do you have?\" to see the full index\n\nWhen you ask about a topic not covered in Core Frameworks below, I will read the relevant\nchapter file before answering.\n\n---\n\n## Core Frameworks & Mental Models\n\n**The problem**: LLMs have a persistent per-step error rate. A 1%-per-step error rate is\nexpected to fail by step ~100. Standard benchmarks (independent examples, averaged accuracy)\nhide this — they don't expose what happens once you chain thousands or millions of *dependent*\nsteps. This paper solves a task requiring over **one million** dependent LLM steps with **zero\nerrors**, and argues the fix is architectural, not \"wait for a smarter model.\"\n\n**MDAP (Massively Decomposed Agentic Process)** — the general paradigm: decompose a task into\nthe smallest possible subtasks, then apply subtask-level error correction. **MAKER** is this\npaper's concrete implementation: **M**aximal **A**gentic decomposition + first-to-ahead-by-**K**\n**E**rror correction + **R**ed-flagging.\n\n**1. Maximal Agentic Decomposition (MAD)** — set subtask size m=1 (one action per LLM call).\nEach agent gets only `(current state, fixed strategy)` — never the accumulated history of past\nactions. This avoids the reliability degradation single agents suffer as their own context\ngrows, and lets you use small, cheap, non-reasoning models. *Use X when Y*: use MAD whenever a\ntask decomposes into a chain\u002Frecursion of minimal, independently-checkable actions.\n\n**2. First-to-ahead-by-k voting** — resample a subtask until one candidate has been chosen k\nmore times than every other candidate. Grounded in the Sequential Probability Ratio Test and\nthe gambler's-ruin problem. Critically, `k_min` (the smallest k hitting a target success\nprobability) grows only **`Θ(ln s)`** — logarithmically in total step count s.\n\n**3. Red-flagging** — discard (don't repair) any response whose structure signals confusion:\noverlong output, or output that fails a *strict* format parser. Its main benefit is suppressing\n**correlated errors** (independent votes colliding on the same wrong answer), not just raising\naverage per-step success rate.\n\n**The scaling-law punchline** (the single most important equation): total expected cost is\n**`Θ(p^-m · c · s · ln s)`** — **exponential in subtask size m**, but only **log-linear in\ntotal step count s**. Corollary: decompose maximally (small m); don't fear long tasks (large\ns). Bundling steps \"for efficiency\" is the single biggest cost mistake this framework warns\nagainst.\n\n**Model\u002Fcost selection rule**: rank candidate models by **`c\u002Fp`** (cost per token ÷ per-step\nsuccess rate), not by raw price or \"reasoning\" reputation. Calibrate p on a small, cheap,\nground-truth-known sample *before* committing to a large-scale run. In this paper's\nexperiments, small non-reasoning models matched or beat reasoning models on cost-effectiveness.\n\n**Verify voting is actually working**: don't trust average error rate alone — explicitly count\n*collisions* (steps whose first two independent votes are both wrong). A low average error rate\ncan still hide dangerous correlated failures that only collision-counting exposes.\n\n**Scope**: this paper covers *execution* (faithfully carrying out a given strategy) — not\n*insight* (generating the strategy\u002Fplan itself). Know which one your problem needs.\n\n---\n\n## Chapter Index\n\n| # | Title | Key Frameworks |\n|---|-------|----------------|\n| [ch01](chapters\u002Fch01-introduction.md) | Introduction — Why Million-Step Tasks Break LLMs | MDAP, MAKER, multi-agent advantage |\n| [ch02](chapters\u002Fch02-background.md) | Background — Error Correction and the Hanoi Testbed | LbAs, voting\u002Fensembling, decomposition granularity |\n| [ch03](chapters\u002Fch03-methods.md) | Methods — MAD, Voting, Red-Flagging | MAD, first-to-ahead-by-k voting, red-flagging, cost\u002Fscaling equations |\n| [ch04](chapters\u002Fch04-experiments.md) | Experiments — The 20-Disk Zero-Error Run | Calibrate-then-scale, model\u002Fcost selection, convergence behavior |\n| [ch05](chapters\u002Fch05-discussion.md) | Red-Flagging Impact, Discussion, Future Work | Collision counting, insight-vs-execution, microservices analogy, safety |\n| [ch06](chapters\u002Fch06-appendix-derivations.md) | Appendix A\u002FB — k_min Derivation | Full derivation of k_min = Θ(ln s) |\n| [ch07](chapters\u002Fch07-appendix-prompts-parsers.md) | Appendix C\u002FD — Prompts, Parsers, Samples | Reusable prompt template, repairing vs. red-flagging parsers |\n\n## Topic Index\n\n- **Calibrate-then-scale workflow** → ch04\n- **Collision counting \u002F correlated errors** → ch05\n- **Cost scaling laws (Θ(s ln s), Θ(p^-m))** → ch03, ch04, ch06\n- **Decomposition granularity (m)** → ch02, ch03\n- **First-to-ahead-by-k voting** → ch03, ch06\n- **Insight vs. execution** → ch05\n- **k_min derivation** → ch03, ch06\n- **MAD (Maximal Agentic Decomposition)** → ch01, ch03\n- **MAKER (the implementation)** → ch01, ch03, ch04\n- **MDAP (the general framework)** → ch01\n- **Microservices analogy** → ch05\n- **Model selection \u002F cost projection** → ch04\n- **Multi-agent advantage** → ch01\n- **Parsers (repairing vs. red-flagging)** → ch03, ch07\n- **Prompt template (Towers of Hanoi agent)** → ch07\n- **Red-flagging** → ch03, ch05, ch07\n- **Safety implications of decomposition** → ch05\n- **Towers of Hanoi benchmark** → ch02, ch04\n- **Voting theory (SPRT, gambler's ruin)** → ch02, ch03\n\n## Supporting Files\n\n- [glossary.md](glossary.md) — all key terms with definitions\n- [patterns.md](patterns.md) — MAD, voting, red-flagging, calibrate-then-scale, collision counting\n- [cheatsheet.md](cheatsheet.md) — decision rules, thresholds, model-selection matrix\n\n---\n\n## Scope & Limits\n\nThis skill covers this single paper's content only (arXiv:2511.09030, v1, Nov 2025). It\naddresses *execution* of long-horizon agentic tasks, not open-ended plan\u002Fstrategy generation\n(\"insight\"). Numeric thresholds cited (e.g. the ~700-token red-flagging cutoff) are specific to\nthis paper's Towers-of-Hanoi setup and prompt design — re-derive them empirically for any other\ndomain rather than reusing them as universal constants. For applying these ideas to your own\ncodebase or agent framework, combine with project-specific tools; for topics beyond this paper,\nask directly.\n","\u003C!-- argument-hint: [topic, framework name, or chapter number] -->\n\n# Solving a Million-Step LLM Task with Zero Errors\n**Authors**: Elliot Meyerson, Giuseppe Paolo, Roberto Dailey, Hormoz Shahrzad, Olivier Francon,\nConor F. Hayes, Xin Qiu, Babak Hodjat, Risto Miikkulainen (Cognizant AI Lab \u002F UT Austin)\n**Source**: [arXiv:2511.09030](https:\u002F\u002Farxiv.org\u002Fabs\u002F2511.09030) (Nov 2025) ([Notice](NOTICE.md)) | **Pages**: ~29 | **Chapters**: 7 | **Generated**: 2026-08-26\n\n## How to Use This Skill\n\n- **Without arguments** — load core frameworks below for reference\n- **With a topic** — ask about `voting`, `red-flagging`, `cost scaling`, `decomposition`, or\n  another indexed topic; I find and read the relevant chapter\n- **With a chapter** — ask for `ch03`; I load that specific chapter\n- **Browse** — ask \"what chapters do you have?\" to see the full index\n\nWhen you ask about a topic not covered in Core Frameworks below, I will read the relevant\nchapter file before answering.\n\n---\n\n## Core Frameworks & Mental Models\n\n**The problem**: LLMs have a persistent per-step error rate. A 1%-per-step error rate is\nexpected to fail by step ~100. Standard benchmarks (independent examples, averaged accuracy)\nhide this — they don't expose what happens once you chain thousands or millions of *dependent*\nsteps. This paper solves a task requiring over **one million** dependent LLM steps with **zero\nerrors**, and argues the fix is architectural, not \"wait for a smarter model.\"\n\n**MDAP (Massively Decomposed Agentic Process)** — the general paradigm: decompose a task into\nthe smallest possible subtasks, then apply subtask-level error correction. **MAKER** is this\npaper's concrete implementation: **M**aximal **A**gentic decomposition + first-to-ahead-by-**K**\n**E**rror correction + **R**ed-flagging.\n\n**1. Maximal Agentic Decomposition (MAD)** — set subtask size m=1 (one action per LLM call).\nEach agent gets only `(current state, fixed strategy)` — never the accumulated history of past\nactions. This avoids the reliability degradation single agents suffer as their own context\ngrows, and lets you use small, cheap, non-reasoning models. *Use X when Y*: use MAD whenever a\ntask decomposes into a chain\u002Frecursion of minimal, independently-checkable actions.\n\n**2. First-to-ahead-by-k voting** — resample a subtask until one candidate has been chosen k\nmore times than every other candidate. Grounded in the Sequential Probability Ratio Test and\nthe gambler's-ruin problem. Critically, `k_min` (the smallest k hitting a target success\nprobability) grows only **`Θ(ln s)`** — logarithmically in total step count s.\n\n**3. Red-flagging** — discard (don't repair) any response whose structure signals confusion:\noverlong output, or output that fails a *strict* format parser. Its main benefit is suppressing\n**correlated errors** (independent votes colliding on the same wrong answer), not just raising\naverage per-step success rate.\n\n**The scaling-law punchline** (the single most important equation): total expected cost is\n**`Θ(p^-m · c · s · ln s)`** — **exponential in subtask size m**, but only **log-linear in\ntotal step count s**. Corollary: decompose maximally (small m); don't fear long tasks (large\ns). Bundling steps \"for efficiency\" is the single biggest cost mistake this framework warns\nagainst.\n\n**Model\u002Fcost selection rule**: rank candidate models by **`c\u002Fp`** (cost per token ÷ per-step\nsuccess rate), not by raw price or \"reasoning\" reputation. Calibrate p on a small, cheap,\nground-truth-known sample *before* committing to a large-scale run. In this paper's\nexperiments, small non-reasoning models matched or beat reasoning models on cost-effectiveness.\n\n**Verify voting is actually working**: don't trust average error rate alone — explicitly count\n*collisions* (steps whose first two independent votes are both wrong). A low average error rate\ncan still hide dangerous correlated failures that only collision-counting exposes.\n\n**Scope**: this paper covers *execution* (faithfully carrying out a given strategy) — not\n*insight* (generating the strategy\u002Fplan itself). Know which one your problem needs.\n\n---\n\n## Chapter Index\n\n| # | Title | Key Frameworks |\n|---|-------|----------------|\n| [ch01](chapters\u002Fch01-introduction.md) | Introduction — Why Million-Step Tasks Break LLMs | MDAP, MAKER, multi-agent advantage |\n| [ch02](chapters\u002Fch02-background.md) | Background — Error Correction and the Hanoi Testbed | LbAs, voting\u002Fensembling, decomposition granularity |\n| [ch03](chapters\u002Fch03-methods.md) | Methods — MAD, Voting, Red-Flagging | MAD, first-to-ahead-by-k voting, red-flagging, cost\u002Fscaling equations |\n| [ch04](chapters\u002Fch04-experiments.md) | Experiments — The 20-Disk Zero-Error Run | Calibrate-then-scale, model\u002Fcost selection, convergence behavior |\n| [ch05](chapters\u002Fch05-discussion.md) | Red-Flagging Impact, Discussion, Future Work | Collision counting, insight-vs-execution, microservices analogy, safety |\n| [ch06](chapters\u002Fch06-appendix-derivations.md) | Appendix A\u002FB — k_min Derivation | Full derivation of k_min = Θ(ln s) |\n| [ch07](chapters\u002Fch07-appendix-prompts-parsers.md) | Appendix C\u002FD — Prompts, Parsers, Samples | Reusable prompt template, repairing vs. red-flagging parsers |\n\n## Topic Index\n\n- **Calibrate-then-scale workflow** → ch04\n- **Collision counting \u002F correlated errors** → ch05\n- **Cost scaling laws (Θ(s ln s), Θ(p^-m))** → ch03, ch04, ch06\n- **Decomposition granularity (m)** → ch02, ch03\n- **First-to-ahead-by-k voting** → ch03, ch06\n- **Insight vs. execution** → ch05\n- **k_min derivation** → ch03, ch06\n- **MAD (Maximal Agentic Decomposition)** → ch01, ch03\n- **MAKER (the implementation)** → ch01, ch03, ch04\n- **MDAP (the general framework)** → ch01\n- **Microservices analogy** → ch05\n- **Model selection \u002F cost projection** → ch04\n- **Multi-agent advantage** → ch01\n- **Parsers (repairing vs. red-flagging)** → ch03, ch07\n- **Prompt template (Towers of Hanoi agent)** → ch07\n- **Red-flagging** → ch03, ch05, ch07\n- **Safety implications of decomposition** → ch05\n- **Towers of Hanoi benchmark** → ch02, ch04\n- **Voting theory (SPRT, gambler's ruin)** → ch02, ch03\n\n## Supporting Files\n\n- [glossary.md](glossary.md) — all key terms with definitions\n- [patterns.md](patterns.md) — MAD, voting, red-flagging, calibrate-then-scale, collision counting\n- [cheatsheet.md](cheatsheet.md) — decision rules, thresholds, model-selection matrix\n\n---\n\n## Scope & Limits\n\nThis skill covers this single paper's content only (arXiv:2511.09030, v1, Nov 2025). It\naddresses *execution* of long-horizon agentic tasks, not open-ended plan\u002Fstrategy generation\n(\"insight\"). Numeric thresholds cited (e.g. the ~700-token red-flagging cutoff) are specific to\nthis paper's Towers-of-Hanoi setup and prompt design — re-derive them empirically for any other\ndomain rather than reusing them as universal constants. For applying these ideas to your own\ncodebase or agent framework, combine with project-specific tools; for topics beyond this paper,\nask directly.\n",{"name":5,"domain":49,"description":7},"engineering","main","cd89f013b617f7eea982a9bbc60e992974fda2aa","skills\u002Fmaker-mdap",15,"https:\u002F\u002Fgithub.com\u002Fcoco-research\u002Fcoco","https:\u002F\u002Fcocoresearch.org\u002Fcoco\u002F","https:\u002F\u002Fgithub.com\u002Fcoco-research","https:\u002F\u002Fcocoresearch.org","Local-first, open-source tools for people who work with AI agents: governed memory, agents that check their own work, and data that stays on your machine.","2026-11-11 12:18:13",{"source":8,"skillName":5,"commands":61,"prompt":71,"zipAvailable":72},[62,65,68],{"tool":63,"command":64},"npx","npx skills add coco-research\u002Fcoco --skill maker-mdap",{"tool":66,"command":67},"bunx","bunx skills add coco-research\u002Fcoco --skill maker-mdap",{"tool":69,"command":70},"pnpm","pnpm dlx skills add coco-research\u002Fcoco --skill maker-mdap","安装技能，技能地址：https:\u002F\u002Fgithub.com\u002Fcoco-research\u002Fcoco\u002Fblob\u002Fmain\u002Fskills\u002Fmaker-mdap\u002FSKILL.md",true,[74,80,84,88,92,96,100,104,108,112,116],{"path":75,"resourceType":76,"resourceTypeDesc":77,"sizeBytes":78,"sourceHtmlUrl":79},"chapters\u002Fch01-introduction.md","REFERENCE","参考文档",3625,"https:\u002F\u002Fgithub.com\u002Fcoco-research\u002Fcoco\u002Fblob\u002Fmain\u002Fskills\u002Fmaker-mdap\u002Fchapters\u002Fch01-introduction.md",{"path":81,"resourceType":76,"resourceTypeDesc":77,"sizeBytes":82,"sourceHtmlUrl":83},"chapters\u002Fch02-background.md",4541,"https:\u002F\u002Fgithub.com\u002Fcoco-research\u002Fcoco\u002Fblob\u002Fmain\u002Fskills\u002Fmaker-mdap\u002Fchapters\u002Fch02-background.md",{"path":85,"resourceType":76,"resourceTypeDesc":77,"sizeBytes":86,"sourceHtmlUrl":87},"chapters\u002Fch03-methods.md",7775,"https:\u002F\u002Fgithub.com\u002Fcoco-research\u002Fcoco\u002Fblob\u002Fmain\u002Fskills\u002Fmaker-mdap\u002Fchapters\u002Fch03-methods.md",{"path":89,"resourceType":76,"resourceTypeDesc":77,"sizeBytes":90,"sourceHtmlUrl":91},"chapters\u002Fch04-experiments.md",5820,"https:\u002F\u002Fgithub.com\u002Fcoco-research\u002Fcoco\u002Fblob\u002Fmain\u002Fskills\u002Fmaker-mdap\u002Fchapters\u002Fch04-experiments.md",{"path":93,"resourceType":76,"resourceTypeDesc":77,"sizeBytes":94,"sourceHtmlUrl":95},"chapters\u002Fch05-discussion.md",6431,"https:\u002F\u002Fgithub.com\u002Fcoco-research\u002Fcoco\u002Fblob\u002Fmain\u002Fskills\u002Fmaker-mdap\u002Fchapters\u002Fch05-discussion.md",{"path":97,"resourceType":76,"resourceTypeDesc":77,"sizeBytes":98,"sourceHtmlUrl":99},"chapters\u002Fch06-appendix-derivations.md",1997,"https:\u002F\u002Fgithub.com\u002Fcoco-research\u002Fcoco\u002Fblob\u002Fmain\u002Fskills\u002Fmaker-mdap\u002Fchapters\u002Fch06-appendix-derivations.md",{"path":101,"resourceType":76,"resourceTypeDesc":77,"sizeBytes":102,"sourceHtmlUrl":103},"chapters\u002Fch07-appendix-prompts-parsers.md",8201,"https:\u002F\u002Fgithub.com\u002Fcoco-research\u002Fcoco\u002Fblob\u002Fmain\u002Fskills\u002Fmaker-mdap\u002Fchapters\u002Fch07-appendix-prompts-parsers.md",{"path":105,"resourceType":76,"resourceTypeDesc":77,"sizeBytes":106,"sourceHtmlUrl":107},"cheatsheet.md",4789,"https:\u002F\u002Fgithub.com\u002Fcoco-research\u002Fcoco\u002Fblob\u002Fmain\u002Fskills\u002Fmaker-mdap\u002Fcheatsheet.md",{"path":109,"resourceType":76,"resourceTypeDesc":77,"sizeBytes":110,"sourceHtmlUrl":111},"glossary.md",4527,"https:\u002F\u002Fgithub.com\u002Fcoco-research\u002Fcoco\u002Fblob\u002Fmain\u002Fskills\u002Fmaker-mdap\u002Fglossary.md",{"path":113,"resourceType":76,"resourceTypeDesc":77,"sizeBytes":114,"sourceHtmlUrl":115},"NOTICE.md",586,"https:\u002F\u002Fgithub.com\u002Fcoco-research\u002Fcoco\u002Fblob\u002Fmain\u002Fskills\u002Fmaker-mdap\u002FNOTICE.md",{"path":117,"resourceType":76,"resourceTypeDesc":77,"sizeBytes":118,"sourceHtmlUrl":119},"patterns.md",3578,"https:\u002F\u002Fgithub.com\u002Fcoco-research\u002Fcoco\u002Fblob\u002Fmain\u002Fskills\u002Fmaker-mdap\u002Fpatterns.md",[121,146,156,174,209,228],{"skillNo":122,"name":123,"nameZh":6,"description":124,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":125,"categories":126,"keywords":128,"publishedAt":44},"01M71KEJXF9KMT0BWCMRSRFA4M","trading-super-intelligence","Trading Super Intelligence Team. Named real-world personas across 8 cells. Built local-first (LM Studio) and validator-gated. Illustrative composites; see DISCLAIMER.md.","https:\u002F\u002Fgithub.com\u002Fcoco-research\u002Fcoco\u002Fblob\u002Fmain\u002Fsystems\u002Fsuperintelligence\u002Ftrading\u002FSKILL.md",[127],{"categoryNo":19,"categoryName":20,"taxonomyType":21,"taxonomyTypeDesc":22},[129,130,131,132,133,134,135,136,137,138,139,140,141,142,143,144,145],"super","trading super intelligence","intelligence","trading","validator","personas","local","composites","studio","team","across","world","first","gated","2026","lm studio","local-first",{"skillNo":147,"name":148,"nameZh":6,"description":149,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":150,"categories":151,"keywords":153,"publishedAt":44},"01M71KEFWS3CMAP7DF1P0FWTDE","strategy-super-intelligence","Strategy Super Intelligence Team. Named real-world personas across 7 cells. Built local-first (LM Studio) and validator-gated. Illustrative composites; see DISCLAIMER.md.","https:\u002F\u002Fgithub.com\u002Fcoco-research\u002Fcoco\u002Fblob\u002Fmain\u002Fsystems\u002Fsuperintelligence\u002Fstrategy\u002FSKILL.md",[152],{"categoryNo":19,"categoryName":20,"taxonomyType":21,"taxonomyTypeDesc":22},[154,129,155,131,133,134,135,136,137,138,139,140,141,142,143,144,145],"strategy","strategy super intelligence",{"skillNo":157,"name":158,"nameZh":6,"description":159,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":160,"categories":161,"keywords":168,"publishedAt":44},"01M71KED56ZEPK2QE84QSQPMT0","risk-compliance-super-intelligence","Risk & Compliance (GRC) Super Intelligence Team. Named real-world personas across 7 cells. Built local-first (LM Studio) and validator-gated. Illustrative composites; see DISCLAIMER.md.","https:\u002F\u002Fgithub.com\u002Fcoco-research\u002Fcoco\u002Fblob\u002Fmain\u002Fsystems\u002Fsuperintelligence\u002Frisk-compliance\u002FSKILL.md",[162,163],{"categoryNo":19,"categoryName":20,"taxonomyType":21,"taxonomyTypeDesc":22},{"categoryNo":164,"categoryName":165,"taxonomyType":166,"taxonomyTypeDesc":167},"01JBSEEDCAT000000000000037","律师与法务","PROFESSION","适用职业",[169,129,170,171,172,131,133,134,135,136,173,137,138,139,140,141,142,143,144,145],"risk","risk compliance super intelligence","compliance","risk compliance","grc",{"skillNo":175,"name":176,"nameZh":6,"description":177,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":178,"categories":179,"keywords":193,"publishedAt":44},"01M71KE9A2FWSC74VA96W1EZSN","product-design-super-intelligence","Your product + design brain trust and decision partner. Named personas across 8 cells spanning product strategy, discovery\u002Fresearch, growth\u002Fmetrics, design foundations\u002Fusability, design leadership\u002Fcraft, UX\u002Fcontent, design systems, and a…","https:\u002F\u002Fgithub.com\u002Fcoco-research\u002Fcoco\u002Fblob\u002Fmain\u002Fsystems\u002Fsuperintelligence\u002Fproduct-design\u002FSKILL.md",[180,183,186,187,190],{"categoryNo":181,"categoryName":182,"taxonomyType":21,"taxonomyTypeDesc":22},"01JBSEEDCAT000000000000015","视觉设计",{"categoryNo":184,"categoryName":185,"taxonomyType":21,"taxonomyTypeDesc":22},"01JBSEEDCAT000000000000021","资料检索与研究",{"categoryNo":19,"categoryName":20,"taxonomyType":21,"taxonomyTypeDesc":22},{"categoryNo":188,"categoryName":189,"taxonomyType":166,"taxonomyTypeDesc":167},"01JBSEEDCAT000000000000031","产品经理",{"categoryNo":191,"categoryName":192,"taxonomyType":166,"taxonomyTypeDesc":167},"01JBSEEDCAT000000000000032","设计师",[194,195,131,196,129,197,198,199,200,201,202,138,139,203,204,143,205,206,207,208],"design","product","product design","product design super intelligence","foundations","product-design","decision","design systems","usability","leadership","ux","brain","craft","metrics","trust",{"skillNo":210,"name":211,"nameZh":6,"description":212,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":213,"categories":214,"keywords":217,"publishedAt":44},"01M71KE6H9FPSD9TECYFBPXZ8Y","legal-ip-super-intelligence","Your legal, IP, and regulatory brain trust. 31 named personas across 7 cells covering patent strategy, tech licensing, regulatory litigation, AI governance, open-source compliance, privacy law, and antitrust.","https:\u002F\u002Fgithub.com\u002Fcoco-research\u002Fcoco\u002Fblob\u002Fmain\u002Fsystems\u002Fsuperintelligence\u002Flegal-ip\u002FSKILL.md",[215,216],{"categoryNo":19,"categoryName":20,"taxonomyType":21,"taxonomyTypeDesc":22},{"categoryNo":164,"categoryName":165,"taxonomyType":166,"taxonomyTypeDesc":167},[218,129,219,131,220,221,208,222,134,223,224,225,154,138,139,226,171,227,143,205],"ip","legal ip super intelligence","legal","sep","tech","litigation","ai governance","open","open-source","governance",{"skillNo":229,"name":230,"nameZh":6,"description":231,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":232,"categories":233,"keywords":235,"publishedAt":44},"01M71KE36QRY610V6YVB9JWRND","healthcare-life-sciences-super-intelligence","Your healthcare and life sciences brain trust. 25 named personas across 7 cells covering drug discovery, medtech, biotech strategy, health policy, genomics, clinical trials, and healthcare economics.","https:\u002F\u002Fgithub.com\u002Fcoco-research\u002Fcoco\u002Fblob\u002Fmain\u002Fsystems\u002Fsuperintelligence\u002Fhealthcare-life-sciences\u002FSKILL.md",[234],{"categoryNo":19,"categoryName":20,"taxonomyType":21,"taxonomyTypeDesc":22},[236,129,237,238,131,221,208,239,240,134,241,242,154,138,139,243,244,245,246,143],"healthcare","healthcare life sciences super intelligence","life","economics","discovery","policy","clinical","clinical trials","health","trials","genomics"]