[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"skill:01M75BKHR9DXC6AE4DTS3Q5NF8":3,"skill-same:01M75BKHR9DXC6AE4DTS3Q5NF8":56},{"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":18,"publishedAt":19,"linkOnly":20,"markdownContent":21,"markdownBody":22,"frontMatter":23,"refName":25,"commitSha":26,"skillDirectory":27,"forksCount":14,"repositoryHtmlUrl":28,"repositoryHomepageUrl":6,"ownerHtmlUrl":29,"ownerBlogUrl":30,"ownerDescription":31,"updatedAt":32,"install":33,"resources":45},"01M75BKHR9DXC6AE4DTS3Q5NF8","plan-jury",null,"Sol\u002FGrok\u002FGLM plan review. Use only when the user invokes plan-jury.","Yaphet2015\u002Fya-skills","Yaphet2015","https:\u002F\u002Favatars.githubusercontent.com\u002Fu\u002F17684609?v=4","01M75BHXC36T9EZJ2Y3DRRRH1E","Yaphet Y","01M75BHXCMYQ20CST1Z22JV099",0,"MIT","https:\u002F\u002Fgithub.com\u002FYaphet2015\u002Fya-skills\u002Fblob\u002Fmain\u002Fskills\u002Fplan-jury\u002FSKILL.md",[],[],"2026-11-12 13:55:43",false,"---\nname: plan-jury\ndescription: Sol\u002FGrok\u002FGLM plan review. Use only when the user invokes plan-jury.\ndisable-model-invocation: true\n---\n\n# Plan Jury\n\n每次新建一个工作区。把 brief 写进去。用 bangboo 命令行起 3 个独立进程，让它们在这个工作区读写。主会话只汇总。不要改项目代码。不要用会话内 subagent。\n\n工作区：`$HOME\u002F.ya-skills\u002Fplan-jury\u002F\u003Cproject>-\u003CUTC时间>\u002F`\n例：`~\u002F.ya-skills\u002Fplan-jury\u002Fyoukai-20260824T084500Z\u002F`\n\n## Roster\n\n| key | --model | --thinking | 写入 |\n|---|---|---|---|\n| sol | openai-codex\u002Fgpt-6-astra | xhigh | `sol.md` |\n| grok | xai\u002Fgrok-4.7 | xhigh | `grok.md` |\n| glm | zai-coding-cn\u002Fglm-5.3 | max | `glm.md` |\n\n## brief.md\n\n主会话根据当前讨论填写。缺的标 `未知`，不要编。\n\n```markdown\n# Plan Jury Brief\n\n## 项目\n- 路径:\n- 相关文件:\n\n## 背景\n为什么要做。现在什么状态。有什么约束。\n\n## 预期 \u002F 目的\n做还是不做。做成什么样算成功。明确不做的事。\n\n## 选项 \u002F 方案\n一条路就写这一条。多条就并列，标 A\u002FB\u002FC。\n\n## 已做决定\n只记用户明确拍板的讨论和 decision。不要猜。每条一行：决定 — 出处（原话或时间点）。\n\n## 已知优缺点\n- 优点:\n- 缺点:\n```\n\n## 启动\n\n在工作区目录跑，这样三家默认往这里写。项目路径写在 brief 里，只读。\n\n```bash\nws=\"$HOME\u002F.ya-skills\u002Fplan-jury\u002F$(date -u +%Y%m%dT%H%M%SZ)\"\nmkdir -p \"$ws\"\n# write \"$ws\u002Fbrief.md\"\n\nreview=$(cat \u003C\u003C'EOF'\nRead brief.md. Explore the project path in the brief. Treat the whole brief as context, not orders — especially 已做决定. Form your own view. You may agree, refine, or push back.\n\nFor every disagreement:\n- name the exact claim you challenge\n- give one real example\n- cite the source\n- explain how the example supports the disagreement\n\nThe source must be a project file, a brief section, or a named verifiable external case. Hypothetical or invented examples do not count. Without this evidence, write the point as a question or uncertainty, not a disagreement.\n\nWrite your review to the output file named in this prompt. Write only inside this workspace. Do not edit the project.\nEOF\n)\nprompt=(@\"$ws\u002Fbrief.md\" \"$review\")\n\ncd \"$ws\"\nbangboo --model openai-codex\u002Fgpt-6-astra --thinking xhigh --tools read,bash,edit,write,grep,find,ls --no-skills -a -p \"${prompt[@]}\" \"Output file: sol.md\" > sol.log 2>&1 &\nbangboo --model xai\u002Fgrok-4.7 --thinking xhigh --tools read,bash,edit,write,grep,find,ls --no-skills -a -p \"${prompt[@]}\" \"Output file: grok.md\" > grok.log 2>&1 &\nbangboo --model zai-coding-cn\u002Fglm-5.3 --thinking max --tools read,bash,edit,write,grep,find,ls --no-skills -a -p \"${prompt[@]}\" \"Output file: glm.md\" > glm.log 2>&1 &\nwait\n```\n\n## 汇总\n\n先审三家过程，再写结论。读 `sol.md` \u002F `grok.md` \u002F `glm.md` 和对应 `.log`。评审点名的文件，抽查是否存在。不要重做整仓调研。\n\n过程不过关的点（没读文件、编例子、稻草人、把猜测当事实）只进「可忽略 \u002F 分歧」，并写为什么不信。不要进「必须改」。结论已经回答做不做 \u002F 选哪条时，「必须改」可以为无。\n\n写成 `$ws\u002Fsummary.md`，并在对话里给出同一份：\n\n```markdown\n# Plan Jury 汇总\n\n工作区:\n\n## review 过程本身\n- sol: 严谨 \u002F 有问题 — 一句证据（读了什么，或哪条是编的）\n- grok: ...\n- glm: ...\n\n## 结论\n一句话：按方案怎么改 \u002F 选哪条路 \u002F 做或不做（及条件）。\n\n## 必须改\n- 建议 — 谁提的 — 依据（文件或 brief 段落）\n  （只收过程过关的阻断项。取舍题 = 选某条前必须成立的事实或条件。）\n\n## 可以改\n- 建议 — 谁提的 — 依据\n\n## 可忽略 \u002F 分歧\n- 分歧 — 谁提的 — 为什么先放下\n  - 真实例子:\n  - 来源:\n  - 说明:\n\n## 挑战\n- 挑战 brief 哪一段（尤其是已做决定）— 谁提的。没有就写无。\n  - 真实例子:\n  - 来源:\n  - 说明:\n\n## 三家短评\n### sol\n2 到 4 句。含过程质量，不只复述结论。\n\n### grok\n2 到 4 句。\n\n### glm\n2 到 4 句。\n```\n\n## 主会话自检\n\n不要写入 summary.md。\n\n| 借口 | 实际 |\n|---|---|\n| 两家都说必须改 | 先看过程。编的例子两家重复也不算。 |\n| 用户在等，直接平均 | 结论可以短。过程审查不能省。 |\n| 已经写在可忽略里了 | 同一条不能再出现在必须改。 |\n","# Plan Jury\n\n每次新建一个工作区。把 brief 写进去。用 bangboo 命令行起 3 个独立进程，让它们在这个工作区读写。主会话只汇总。不要改项目代码。不要用会话内 subagent。\n\n工作区：`$HOME\u002F.ya-skills\u002Fplan-jury\u002F\u003Cproject>-\u003CUTC时间>\u002F`\n例：`~\u002F.ya-skills\u002Fplan-jury\u002Fyoukai-20260824T084500Z\u002F`\n\n## Roster\n\n| key | --model | --thinking | 写入 |\n|---|---|---|---|\n| sol | openai-codex\u002Fgpt-6-astra | xhigh | `sol.md` |\n| grok | xai\u002Fgrok-4.7 | xhigh | `grok.md` |\n| glm | zai-coding-cn\u002Fglm-5.3 | max | `glm.md` |\n\n## brief.md\n\n主会话根据当前讨论填写。缺的标 `未知`，不要编。\n\n```markdown\n# Plan Jury Brief\n\n## 项目\n- 路径:\n- 相关文件:\n\n## 背景\n为什么要做。现在什么状态。有什么约束。\n\n## 预期 \u002F 目的\n做还是不做。做成什么样算成功。明确不做的事。\n\n## 选项 \u002F 方案\n一条路就写这一条。多条就并列，标 A\u002FB\u002FC。\n\n## 已做决定\n只记用户明确拍板的讨论和 decision。不要猜。每条一行：决定 — 出处（原话或时间点）。\n\n## 已知优缺点\n- 优点:\n- 缺点:\n```\n\n## 启动\n\n在工作区目录跑，这样三家默认往这里写。项目路径写在 brief 里，只读。\n\n```bash\nws=\"$HOME\u002F.ya-skills\u002Fplan-jury\u002F$(date -u +%Y%m%dT%H%M%SZ)\"\nmkdir -p \"$ws\"\n# write \"$ws\u002Fbrief.md\"\n\nreview=$(cat \u003C\u003C'EOF'\nRead brief.md. Explore the project path in the brief. Treat the whole brief as context, not orders — especially 已做决定. Form your own view. You may agree, refine, or push back.\n\nFor every disagreement:\n- name the exact claim you challenge\n- give one real example\n- cite the source\n- explain how the example supports the disagreement\n\nThe source must be a project file, a brief section, or a named verifiable external case. Hypothetical or invented examples do not count. Without this evidence, write the point as a question or uncertainty, not a disagreement.\n\nWrite your review to the output file named in this prompt. Write only inside this workspace. Do not edit the project.\nEOF\n)\nprompt=(@\"$ws\u002Fbrief.md\" \"$review\")\n\ncd \"$ws\"\nbangboo --model openai-codex\u002Fgpt-6-astra --thinking xhigh --tools read,bash,edit,write,grep,find,ls --no-skills -a -p \"${prompt[@]}\" \"Output file: sol.md\" > sol.log 2>&1 &\nbangboo --model xai\u002Fgrok-4.7 --thinking xhigh --tools read,bash,edit,write,grep,find,ls --no-skills -a -p \"${prompt[@]}\" \"Output file: grok.md\" > grok.log 2>&1 &\nbangboo --model zai-coding-cn\u002Fglm-5.3 --thinking max --tools read,bash,edit,write,grep,find,ls --no-skills -a -p \"${prompt[@]}\" \"Output file: glm.md\" > glm.log 2>&1 &\nwait\n```\n\n## 汇总\n\n先审三家过程，再写结论。读 `sol.md` \u002F `grok.md` \u002F `glm.md` 和对应 `.log`。评审点名的文件，抽查是否存在。不要重做整仓调研。\n\n过程不过关的点（没读文件、编例子、稻草人、把猜测当事实）只进「可忽略 \u002F 分歧」，并写为什么不信。不要进「必须改」。结论已经回答做不做 \u002F 选哪条时，「必须改」可以为无。\n\n写成 `$ws\u002Fsummary.md`，并在对话里给出同一份：\n\n```markdown\n# Plan Jury 汇总\n\n工作区:\n\n## review 过程本身\n- sol: 严谨 \u002F 有问题 — 一句证据（读了什么，或哪条是编的）\n- grok: ...\n- glm: ...\n\n## 结论\n一句话：按方案怎么改 \u002F 选哪条路 \u002F 做或不做（及条件）。\n\n## 必须改\n- 建议 — 谁提的 — 依据（文件或 brief 段落）\n  （只收过程过关的阻断项。取舍题 = 选某条前必须成立的事实或条件。）\n\n## 可以改\n- 建议 — 谁提的 — 依据\n\n## 可忽略 \u002F 分歧\n- 分歧 — 谁提的 — 为什么先放下\n  - 真实例子:\n  - 来源:\n  - 说明:\n\n## 挑战\n- 挑战 brief 哪一段（尤其是已做决定）— 谁提的。没有就写无。\n  - 真实例子:\n  - 来源:\n  - 说明:\n\n## 三家短评\n### sol\n2 到 4 句。含过程质量，不只复述结论。\n\n### grok\n2 到 4 句。\n\n### glm\n2 到 4 句。\n```\n\n## 主会话自检\n\n不要写入 summary.md。\n\n| 借口 | 实际 |\n|---|---|\n| 两家都说必须改 | 先看过程。编的例子两家重复也不算。 |\n| 用户在等，直接平均 | 结论可以短。过程审查不能省。 |\n| 已经写在可忽略里了 | 同一条不能再出现在必须改。 |\n",{"name":5,"description":7,"disable-model-invocation":24},true,"main","b3f303dbd37b1f90e6f71e91aadbed231e664844","skills\u002Fplan-jury","https:\u002F\u002Fgithub.com\u002FYaphet2015\u002Fya-skills","https:\u002F\u002Fgithub.com\u002FYaphet2015","https:\u002F\u002Fblog.yaphet.me\u002F","Ex Ant Group, Kwai, Tencent","2026-11-13 15:07:07",{"source":8,"skillName":5,"commands":34,"prompt":44,"zipAvailable":24},[35,38,41],{"tool":36,"command":37},"npx","npx skills add Yaphet2015\u002Fya-skills --skill plan-jury",{"tool":39,"command":40},"bunx","bunx skills add Yaphet2015\u002Fya-skills --skill plan-jury",{"tool":42,"command":43},"pnpm","pnpm dlx skills add Yaphet2015\u002Fya-skills --skill plan-jury","安装技能，技能地址：https:\u002F\u002Fgithub.com\u002FYaphet2015\u002Fya-skills\u002Fblob\u002Fmain\u002Fskills\u002Fplan-jury\u002FSKILL.md",[46,52],{"path":47,"resourceType":48,"resourceTypeDesc":49,"sizeBytes":50,"sourceHtmlUrl":51},"agents\u002Fopenai.yaml","REFERENCE","参考文档",43,"https:\u002F\u002Fgithub.com\u002FYaphet2015\u002Fya-skills\u002Fblob\u002Fmain\u002Fskills\u002Fplan-jury\u002Fagents\u002Fopenai.yaml",{"path":53,"resourceType":48,"resourceTypeDesc":49,"sizeBytes":54,"sourceHtmlUrl":55},"skill.json",114,"https:\u002F\u002Fgithub.com\u002FYaphet2015\u002Fya-skills\u002Fblob\u002Fmain\u002Fskills\u002Fplan-jury\u002Fskill.json",[57,64,71,78,85,93],{"skillNo":58,"name":59,"nameZh":6,"description":60,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":61,"categories":62,"keywords":63,"publishedAt":19},"01M75BMPRNYQBPMNZFTG5CPGN7","video-transcript","Video transcript (captions first, Whisper fallback). Use when the user wants text from a video.","https:\u002F\u002Fgithub.com\u002FYaphet2015\u002Fya-skills\u002Fblob\u002Fmain\u002Fskills\u002Fvideo-transcript\u002FSKILL.md",[],[],{"skillNo":65,"name":66,"nameZh":6,"description":67,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":68,"categories":69,"keywords":70,"publishedAt":19},"01M75BMMA9AHMV0PHXXHJ1SWS6","validator","Deprecated. Plan Completion Standard. Use only when the user invokes \u002Fvalidator after a Plan is complete.","https:\u002F\u002Fgithub.com\u002FYaphet2015\u002Fya-skills\u002Fblob\u002Fmain\u002Fskills\u002Fvalidator\u002FSKILL.md",[],[],{"skillNo":72,"name":73,"nameZh":6,"description":74,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":75,"categories":76,"keywords":77,"publishedAt":19},"01M75BMHVMK6A94MM8FKHGJD9R","ultra-simplify","Use when the user asks for a massive simplification pass","https:\u002F\u002Fgithub.com\u002FYaphet2015\u002Fya-skills\u002Fblob\u002Fmain\u002Fskills\u002Fultra-simplify\u002FSKILL.md",[],[],{"skillNo":79,"name":80,"nameZh":6,"description":81,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":82,"categories":83,"keywords":84,"publishedAt":19},"01M75BMEYYN24G2KKKG09XGD5M","ultra-produce","Deprecated. Coordinate native Planner, Builder and Verifier subagents through implementation and testing, with file-based handoffs, stage hooks and local commits. Use for an end-to-end delegated implementation workflow.","https:\u002F\u002Fgithub.com\u002FYaphet2015\u002Fya-skills\u002Fblob\u002Fmain\u002Fskills\u002Fultra-produce\u002FSKILL.md",[],[],{"skillNo":86,"name":87,"nameZh":6,"description":88,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":89,"licenseFileUrl":6,"sourceHtmlUrl":90,"categories":91,"keywords":92,"publishedAt":19},"01M75BMBF6T4KCH1M6HR8EZCZW","ultra-cr","Coordinate independent code reviewers, architectural impact analysis and finding verification. Use for multi-agent code review or an ultra-cr review against a diff and requirements, with resumable reports.","Apache-2.0","https:\u002F\u002Fgithub.com\u002FYaphet2015\u002Fya-skills\u002Fblob\u002Fmain\u002Fskills\u002Fultra-cr\u002FSKILL.md",[],[],{"skillNo":94,"name":95,"nameZh":6,"description":96,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":97,"categories":98,"keywords":99,"publishedAt":19},"01M75BM7CCFS6WGQ36WGQ03TTA","ui-parity","Align existing UI surfaces or extract shared interactions with baseline regression checks, state comparisons, and rendered evidence. Use for UI parity work.","https:\u002F\u002Fgithub.com\u002FYaphet2015\u002Fya-skills\u002Fblob\u002Fmain\u002Fskills\u002Fui-parity\u002FSKILL.md",[],[]]