agentsclimarketplace

Dispatch rules

Skill cheukswiss/etdxm/.claude/skills/dispatch-rules

尚书省任务派发规则与调度策略,定义六部派发原则、跨部门协作与并行串行策略From its SKILL.md

Install
npx -y skills add cheukswiss/etdxm --skill dispatch-rules

Assembled from the repository path, not quoted from the project. Check it against their README if it does not work.

One thing to look at

  • 2 stars2 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.

SKILL.md

2.9 KB, ~1.0k tokens by cl100k_base, as published. Nobody here has run it

Skill: 任务派发规则与调度策略

适用 Agent:尚书省 加载时机:尚书省 dispatch 模式(收到准奏方案准备派发时)

派发原则

任务类型派发至示例
功能开发、代码实现、Bug 修复🔧 工部实现登录接口、修复分页 Bug
数据库设计、数据处理、报表💰 户部设计用户表、生成月报
文档编写、规范制定📝 礼部编写 API 文档、制定 Git 规范
部署、CI/CD、服务器运维⚔️ 兵部配置 GitHub Actions、部署到 K8s
测试、审计、安全扫描⚖️ 刑部编写单元测试、OWASP 扫描
Agent 管理、权限配置📋 吏部添加新 Agent、调整权限

跨部门协作

当一个子任务涉及多个部门时:

  1. 拆分:将复合任务拆为独立子任务,明确每部分负责部门
  2. 定义依赖:标注执行顺序(如工部先开发 → 礼部写文档 → 刑部写测试)
  3. 接口约定:定义各部门之间的数据传递格式
  4. 同步机制:前序任务完成后,通知后续部门启动

调度策略

  1. 依赖优先:先派发无依赖的任务,再按依赖链派发后续任务
  2. 并行最大化:无依赖关系的任务同时派发,缩短总执行时间
  3. 关键路径:识别最长执行链,优先保障关键路径上的任务
  4. 负载均衡:避免单一部门堆积过多任务
  5. 快速反馈:短任务优先完成,让皇上尽早看到进展

📎 容错与降级策略详见 fault-tolerance-tables.md

部门内并行提示(parallel_hint)

尚书省派发时,若同一部门收到多个属于同一 parallel_group 的任务,须进一步判断并在工单中附加 parallel_hint

{
  "task": {
    "id": "TASK-001",
    "assigned_to": "gongbu",
    "parallel_hint": {
      "can_parallel_with": ["TASK-005"],
      "reason": "T-001 修改 src/auth/,T-005 修改 src/api/,模块不重叠",
      "confidence": "high"
    }
  }
}
  • can_parallel_with:可与哪些同部门任务并行
  • reason:并行判断的理由
  • confidencehigh(确定无冲突)/ medium(大概率无冲突,堂官需验证)/ low(不确定,堂官自行判断)
  • 分析粒度:模块级——同部门任务影响的模块是否重叠

注意parallel_hint 是建议而非命令,堂官有权基于文件级冲突检测推翻。

📎 堂官收到 parallel_hint 后的冲突检测规则详见 parallel-conflict-rules.md

并行与串行示例

并行:工部(前端) + 户部(数据库) + 礼部(文档) ──→ 尚书汇总 → 刑部集成测试
串行:工部(核心逻辑) → 刑部(测试) → 兵部(部署)

What ships with it

Read from the repository

Just SKILL.md. No reference files, no scripts.

Keep looking

Skills are one crate of 325,949. Ordering is by how many stacks a row turns up in, so the top of any crate is what has actually been picked rather than what has the most stars.