router
Skill2 starsChanged 13 days agoMITView on GitHub
派发与路由调度。决定当前阶段该用哪个技能,以及架构师派活时必须把哪几样东西装进派发包——保证角色信息、边界、判据、证据落点不丢。触发词:路由/调度/该用什么skill/派活/派发/分工/接手项目/交接/编排/该谁做。
18 other projects use a version of this.
Adopted by
Most recently taken up by
- dcamunas/model-routertheir own version, first seen Nov 15, 2026
- grainulation/grainulatortheir own version, first seen Nov 12, 2026
- gamedev-skills/awesome-gamedev-agent-skillstheir own version, first seen Nov 11, 2026
- lizzyDeng/claude-skillstheir own version, first seen Nov 8, 2026
- apattichis/agentic-research-toolkittheir own version, first seen Nov 8, 2026
- Wellsonr/skillsettheir own version, first seen Nov 8, 2026
Used with
Projects with it also use setup.
Other files people use alongside this one.
Files found in the same public projects as a version of this one, more often than they turn up anywhere else. Each project counts once, large collections count for less, and this file's own repository is left out.
VercelSkillsetup
Set up Vercel CLI and project configuration. Use when the user says "set up Vercel", "configure Vercel", "link to Vercel", or "vercel init".
Also in 3 projects that use this
Other versions18 projects use a version of this.
Word for word · 0
None found.
Edited · 0
None found.
Wrote their own version · 18
- gamedev-skills/awesome-gamedev-agent-skills★ 9ksame name
- grainulation/grainulator★ 86same name
- lubusIN/frappe-skills★ 68same name
- timharris707/skills★ 44same name
- jammyfu/open-game-skills★ 12same name
- Canary-ECAD/eplan-scripting-skill★ 5same name
- Aetherdz/huntpack★ 4same name
- SunCherry/agentic-AI_analog_IC_auto-designer★ 4same name
- Adnova-Group/muster★ 2same name
- letsloose501/sqs-agent-memory★ 2same name
8 more not listed
How matches are found. Word for word: the exact same text. Edited: shares passages with this file (“99% same” means 99% of its five-word phrases also appear here; 10% or more counts). Their own version: similar meaning, or, weakest, the same name. We can’t tell which project had it first. Projects are counted among the 200 most-starred candidates, and mentions are links to the repository we found, up to 50 per site, so both counts are a floor.
In this repository15 other files it gives its agents.
- skills/iso/light/CLAUDE.mdCLAUDE.md
- ai-code-reviewSkill
- api-design-reviewSkill
- architecture-support-briefSkill
- class-design-reviewSkill
- comment-policySkill
- cpp-exemplarsSkill
- error-code-classificationSkill
- essential-tool-installSkill
- isoSkill
- notion-diagramSkill
- reportSkill
- researchSkill
- incident-reviewSkill
- test-naming-and-structureSkill
The file
Files
派发调度器
两件事:选技能,派任务。数据源是 catalog.json(由 scaffold/skill-lint.py --write 生成,不手改);本文件是决策。
0. 四条最高准则(全文在 scaffold/policies/top-principles.md)
| 准则 | 一句话 |
| Reuse First | 能在现有实现上改就别另起一套;偏离留 DV,无记录的绕开=缺陷 |
| Global Over Local | 局部收益放进系统边界称重:调用方清单、契约 diff、受影响门禁 |
| Falsifiability | 什么现象算错 / 尺子在基线上给出别的值吗 / 把实现改坏哪个用例会红 |
| Single Source of Truth | 一处定义、他处引用;复制即分叉 |
派活时不需要复述全文,但派发包里出现与准则冲突的要求(另起一套、只测一条路径、两处写同一规则)要被指出来——冲突是派发方的缺陷。
1. 阶段 → 技能
| 判定(用户说了什么) | 技能 |
| 项目启动 / 分析需求 / 做架构 / 拆子系统 / 实现 / 集成 / 准备上线 | iso |
| 代码审查、finding、评审收口 | ai-code-review |
| 写用例、评审用例命名与结构 | test-naming-and-structure |
| 注释该留不该留 | comment-policy |
| 返回码 / 错误模型 / 观测面分工 | error-code-classification |
| 画架构图 / 分层图 / 流程图(SVG) | notion-diagram |
| 写报告、方案答辩包 | report / architecture-support-brief |
| 故障复盘、postmortem | review/incident-review |
| 缺工具、需要安装权限 | essential-tool-install |
| 写新类前要对样本 | cpp-exemplars |
| 门禁、台账、提交、角色边界、停线 | scaffold/sop + scaffold/policies(不是技能) |
判据(反写):两个技能都能接住同一个请求时,先看谁给的是判据、谁给的是方法——判据优先。仍分不出来,说明触发词该收窄,改 description,不留双重入口。
2. 派发:一包五样,缺项不得开工
先判该不该派,再判包齐不齐。只有两种需求买得到角色(判据与成本见 sop/collaboration-flow.md §1):独立性(要它推翻某个结论,同时答得出「推翻什么」)与并行(去掉它主线会停)。其余请求自己执行、按 §1 选技能——派出去要重述上下文、回来要复核产出,换不到这两样就是净亏。
架构师派活时不发实现步骤,发一张派发包——就是 scaffold/templates/task-card.tpl 的前三节(角色与必装技能 / 目标与边界 / 验收判据与证据),落盘到 <日期>/<卡号>/卡.md(卡目录布局见 sop/collaboration-flow.md §9.1)。派发包与任务卡是同一件东西的两种叫法,不另立模板。
为什么落盘:对话会被压缩、会被新任务挤出,文件不会。角色信息只存在于对话里,等于没派。
| 槽位 | 内容 | 缺了会怎样 |
| 角色与边界 | 谁做、不许碰什么、谁复核 | 执行者扩大范围,或不敢动该动的 |
| 目标 | 做完之后调用方 / 用户能观察到什么变化 | 目标被替换成"跑通测试" |
| 验收判据 | 可判定条件:阈值 + 测量方法 + 失败长什么样 | 交付"看起来完成了",无人能否证 |
| 证据落点 | run 目录 / 用例名 / 报告路径 | 结论不可复核,等于口头 |
| 必装技能 | 按 §3 角色表取,装载后回执写明装了哪几个 | 规范没进执行现场 |
判据(反写):收到派发包的人如果还要回头问「我该做什么、做到什么算完、证据放哪」,说明包缺项——缺项是派发方的缺陷,不是执行方的问题。
3. 角色 → 必备技能
| 角色 | 必装 | 派发时必须交代 |
| 架构师(常驻) | router + iso + 全部 policies;亲自实现时另装 cpp-exemplars、comment-policy、test-naming-and-structure(编码样本与注释纪律不在 policies 里) | 裁决权、唯一写状态文件、自己实现的卡不得自行充当独立评审 |
| Dev | iso(operations/cpp.md)、comment-policy、test-naming-and-structure、error-code-classification | 文件范围、契约、禁止项、门禁命令、提交格式。上下文已在架构师手里时不派 |
| QA | ai-code-review | 评审对象的 commit、独立用例、finding schema、不许复用 Dev 的断言当唯一证据。这是「独立性」的主承担者,不可省 |
| 复盘 owner | review/incident-review | 证据包路径、三层根因要求 |
并行工作树:技能装在 <项目>/.claude/skills 不会随 worktree 生成。用 install-skills.sh --worktrees 或 --user,否则执行现场没有规范,派发包里的"必装技能"是空话。
准则自检(按角色,各一条):架构师 → Reuse First;Dev → Falsifiability;QA → Falsifiability 的「基线」那条;复盘 / 测试 owner → Single Source of Truth。问句原文在 top-principles.md 各条「自检」,派发时不复制。
4. 规则冲突
同一条规则出现在两处 → 以更具体的为准(项目 > 脚手架,判据 > 方法),并把冲突记一行偏离。
不允许两处都留着——留着就是让下一个人靠猜。
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

