[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"skill:01M7DMH7VJ55VC39J4K0SNXSA1":3,"skill-same:01M7DMH7VJ55VC39J4K0SNXSA1":45},{"skillNo":4,"name":5,"nameZh":6,"description":7,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":15,"categories":16,"keywords":17,"publishedAt":18,"linkOnly":19,"markdownContent":20,"markdownBody":21,"frontMatter":22,"refName":24,"commitSha":25,"skillDirectory":26,"forksCount":27,"repositoryHtmlUrl":28,"repositoryHomepageUrl":29,"ownerHtmlUrl":30,"ownerBlogUrl":6,"ownerDescription":6,"updatedAt":31,"install":32,"resources":44},"01M7DMH7VJ55VC39J4K0SNXSA1","swarm",null,"扇出 N 个并行 worker，收齐后返回一份报告。用于 \u002Fswarm、「swarm this」，或并行覆盖、race、gauntlet 与探索。","fenfenxu\u002Fpstack-zh","fenfenxu","https:\u002F\u002Favatars.githubusercontent.com\u002Fu\u002F2194243?v=4","01M7DMDSWBASX3NXC18FPHDJ8G","01M7DMDSWV8ASCZ4PH6YFXHPSJ",3,"MIT","https:\u002F\u002Fgithub.com\u002Ffenfenxu\u002Fpstack-zh\u002Fblob\u002Fmain\u002Fskills\u002Fswarm\u002FSKILL.md",[],[],"2026-11-15 14:19:17",false,"---\nname: swarm\ndescription: \"扇出 N 个并行 worker，收齐后返回一份报告。用于 \u002Fswarm、「swarm this」，或并行覆盖、race、gauntlet 与探索。\"\ndisable-model-invocation: true\n---\n\n# Swarm\n\n## 开始\n\n启动任何东西前先打开 todolist，每阶段一条。\n\n1. 定框\n2. 扇出\n3. 聚合\n4. 报告\n\n## 阶段 A：定框\n\n1. 说明完成谓词与 swarm 须返回的 artifact 或报告。\n2. 选形态。分 slice、N worker 同 brief race，或混合。race 或混合形态须在 spawn 前声明 `first pass`、`rank all` 或 `best-of`。\n4. worker 模型取自 `~\u002F.cursor\u002Frules\u002Fpstack-models.mdc` 的 `swarm workers` 行。规则或该行缺失时用 `grok-4.7-xhigh-fast`。`auto` 或 `inherit-parent` 则省略 `model`，worker 跑在父模型上。Task 拒绝 slug 则用默认并说明。默认也被拒则从错误信息取同族最接近有效 slug。model race 须 upfront 命名各 arm 模型。按这项工作给每个子代理选推理强度，以及用 fast 还是标准速度，见 **poteto-mode** skill 的 Subagents 一节。\n5. 写输出的 worker 各用可写输出路径。worker 验证或度量 commit 时，各 brief 命名确切 SHA。度量 brief 还命名方法（样本数、一样本是什么、顺序）。worker 在结果中记录二者。\n\n## 阶段 B：扇出\n\n一条消息 spawn 全部 N worker：`subagent_type: generalPurpose`、`environment: \"cloud\"`、`run_in_background: true`、步骤 4 的 model（`auto` 或 `inherit-parent` 时不设）。仅当 worker 需访问用户电脑上的东西时用 `environment: \"local\"`。\n\nworker 须从非默认已 push 分支开始时，传 `cloud_base_branch`。\n\n每条 brief 自洽。含目标、范围、确切 slice 或 race arm、如何验证、报告什么。报告用 `PASS`、`ISSUES` 或 `BLOCKED` 加证据。能证明缺陷的 worker 报 `ISSUES` 并列出能证明的全部 issue，不只第一个。\n\n某 worker dropout，以 N-1 继续并注明。\n\n## 阶段 C：聚合\n\n读 terminal 结果。未记录 brief 要求的 SHA 与方法的结果丢弃，重新启动一个新 worker 一次。第二次仍 miss 则记 gap。gap 不算 pass。覆盖形态下每个必需 slice 都要有结果。race 按 upfront 声明的选择规则：`first pass`、`rank all` 或 `best-of`。不要粘贴 raw worker dump。\n\n## 阶段 D：报告\n\n在 chat 内返回一份合并报告：表、issue 一行摘要、gap 或 dropout、若用了 race 规则则写明。\n","# Swarm\n\n## 开始\n\n启动任何东西前先打开 todolist，每阶段一条。\n\n1. 定框\n2. 扇出\n3. 聚合\n4. 报告\n\n## 阶段 A：定框\n\n1. 说明完成谓词与 swarm 须返回的 artifact 或报告。\n2. 选形态。分 slice、N worker 同 brief race，或混合。race 或混合形态须在 spawn 前声明 `first pass`、`rank all` 或 `best-of`。\n4. worker 模型取自 `~\u002F.cursor\u002Frules\u002Fpstack-models.mdc` 的 `swarm workers` 行。规则或该行缺失时用 `grok-4.7-xhigh-fast`。`auto` 或 `inherit-parent` 则省略 `model`，worker 跑在父模型上。Task 拒绝 slug 则用默认并说明。默认也被拒则从错误信息取同族最接近有效 slug。model race 须 upfront 命名各 arm 模型。按这项工作给每个子代理选推理强度，以及用 fast 还是标准速度，见 **poteto-mode** skill 的 Subagents 一节。\n5. 写输出的 worker 各用可写输出路径。worker 验证或度量 commit 时，各 brief 命名确切 SHA。度量 brief 还命名方法（样本数、一样本是什么、顺序）。worker 在结果中记录二者。\n\n## 阶段 B：扇出\n\n一条消息 spawn 全部 N worker：`subagent_type: generalPurpose`、`environment: \"cloud\"`、`run_in_background: true`、步骤 4 的 model（`auto` 或 `inherit-parent` 时不设）。仅当 worker 需访问用户电脑上的东西时用 `environment: \"local\"`。\n\nworker 须从非默认已 push 分支开始时，传 `cloud_base_branch`。\n\n每条 brief 自洽。含目标、范围、确切 slice 或 race arm、如何验证、报告什么。报告用 `PASS`、`ISSUES` 或 `BLOCKED` 加证据。能证明缺陷的 worker 报 `ISSUES` 并列出能证明的全部 issue，不只第一个。\n\n某 worker dropout，以 N-1 继续并注明。\n\n## 阶段 C：聚合\n\n读 terminal 结果。未记录 brief 要求的 SHA 与方法的结果丢弃，重新启动一个新 worker 一次。第二次仍 miss 则记 gap。gap 不算 pass。覆盖形态下每个必需 slice 都要有结果。race 按 upfront 声明的选择规则：`first pass`、`rank all` 或 `best-of`。不要粘贴 raw worker dump。\n\n## 阶段 D：报告\n\n在 chat 内返回一份合并报告：表、issue 一行摘要、gap 或 dropout、若用了 race 规则则写明。\n",{"name":5,"description":7,"disable-model-invocation":23},true,"main","e4477188d43269545432a83e4ac5d0733a812d09","skills\u002Fswarm",0,"https:\u002F\u002Fgithub.com\u002Ffenfenxu\u002Fpstack-zh","http:\u002F\u002Fpstack.ganhai.cloud","https:\u002F\u002Fgithub.com\u002Ffenfenxu","2026-10-21 12:14:41",{"source":8,"skillName":5,"commands":33,"prompt":43,"zipAvailable":23},[34,37,40],{"tool":35,"command":36},"npx","npx skills add fenfenxu\u002Fpstack-zh --skill swarm",{"tool":38,"command":39},"bunx","bunx skills add fenfenxu\u002Fpstack-zh --skill swarm",{"tool":41,"command":42},"pnpm","pnpm dlx skills add fenfenxu\u002Fpstack-zh --skill swarm","安装技能，技能地址：https:\u002F\u002Fgithub.com\u002Ffenfenxu\u002Fpstack-zh\u002Fblob\u002Fmain\u002Fskills\u002Fswarm\u002FSKILL.md",[],[46,53,60,67,74,81],{"skillNo":47,"name":48,"nameZh":6,"description":49,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":50,"categories":51,"keywords":52,"publishedAt":18},"01M7DMHJV7QMEE1F43FFW98FV8","why","用于「为什么 X 是这样实现的」「为什么选了 Y」、设计理由、回归、事后复盘，或数据支撑的阈值问题。发现可用 MCP 并并行查询各证据类别（源码控制、工单跟踪、长文文档、实时聊天、基础设施可观测性、错误跟踪、产品分析数仓），然后返回带引用的决策与权衡解读。运行时行为请用 how。","https:\u002F\u002Fgithub.com\u002Ffenfenxu\u002Fpstack-zh\u002Fblob\u002Fmain\u002Fskills\u002Fwhy\u002FSKILL.md",[],[],{"skillNo":54,"name":55,"nameZh":6,"description":56,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":57,"categories":58,"keywords":59,"publishedAt":18},"01M7DMHHRSS2SCZN6HWSANBJ35","unslop","从任意写作中剔除 AI 痕迹。","https:\u002F\u002Fgithub.com\u002Ffenfenxu\u002Fpstack-zh\u002Fblob\u002Fmain\u002Fskills\u002Funslop\u002FSKILL.md",[],[],{"skillNo":61,"name":62,"nameZh":6,"description":63,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":64,"categories":65,"keywords":66,"publishedAt":18},"01M7DMHGFGH2YT86WQ6A53693W","typescript-best-practices","TypeScript 最佳实践。读取或编辑任意 .ts 或 .tsx 文件时使用。","https:\u002F\u002Fgithub.com\u002Ffenfenxu\u002Fpstack-zh\u002Fblob\u002Fmain\u002Fskills\u002Ftypescript-best-practices\u002FSKILL.md",[],[],{"skillNo":68,"name":69,"nameZh":6,"description":70,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":71,"categories":72,"keywords":73,"publishedAt":18},"01M7DMHEA1EGG2D9XQ33N66GDS","technical-writing","分层技术写作标准：Diátaxis 结构、Google 开发者风格句子、STE 指令规则、Global English 句法。用于 \u002Ftechnical-writing，或撰写\u002F审查文档、RFC、README、PR 描述、commit message 时。","https:\u002F\u002Fgithub.com\u002Ffenfenxu\u002Fpstack-zh\u002Fblob\u002Fmain\u002Fskills\u002Ftechnical-writing\u002FSKILL.md",[],[],{"skillNo":75,"name":76,"nameZh":6,"description":77,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":78,"categories":79,"keywords":80,"publishedAt":18},"01M7DMHCEXE1YRDR6XC1RXBG76","teach","用平实语言解释一段工作，让人真正理解。运行 how 与 why skill，把发现织成一份清晰说明。用于「teach me this」「help me really understand X」「explain this change or subsystem to me」。","https:\u002F\u002Fgithub.com\u002Ffenfenxu\u002Fpstack-zh\u002Fblob\u002Fmain\u002Fskills\u002Fteach\u002FSKILL.md",[],[],{"skillNo":82,"name":83,"nameZh":6,"description":84,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":85,"categories":86,"keywords":87,"publishedAt":18},"01M7DMH9KNBP83SHAW4TGZ4713","tdd","仅当用户明确要求 TDD、失败测试或回归测试时使用，或 bug 有明显且廉价的本地测试目标时使用。测试路径不清、昂贵、偏集成或未要求时跳过。","https:\u002F\u002Fgithub.com\u002Ffenfenxu\u002Fpstack-zh\u002Fblob\u002Fmain\u002Fskills\u002Ftdd\u002FSKILL.md",[],[]]