
Team Coordinator
- 34 installs
- 82 repo stars
- Updated August 2, 2026
- aaaaqwq/claude-code-skills
team-coordinator is a Claude Code skill that makes the agent act as an executive, decomposing a task and routing sub-tasks to specialist worker agents.
About
team-coordinator makes the agent act as an executive that breaks a user task down and assigns it to the most suitable specialist worker agents, coordinating parallel work and reviewing the output. A developer uses it to route complex, multi-domain requests across a team of agents instead of doing the work directly. It defines a capability matrix and a task-routing decision flow.
- Acts as an executive that decomposes a task and routes it to specialist agents
- Defines a capability matrix of eight worker agents by strength and fit
- Coordinates parallel multi-agent work and reviews the combined output
Team Coordinator by the numbers
- 34 all-time installs (skills.sh)
- Ranked #8,777 of 16,556 AI & Agent Building skills by installs in the Skillselion catalog
- Data as of Aug 3, 2026 (Skillselion catalog sync)
team-coordinator capabilities & compatibility
- Capabilities
- agent orchestration · task routing · task decomposition
- Use cases
- orchestration · planning · project management
- Pricing
- Free
What team-coordinator says it does
sessions_spawn(task="...", agentId="code")
license: MIT
npx skills add https://github.com/aaaaqwq/claude-code-skills --skill team-coordinatorAdd your badge
Show developers this skill is listed on Skillselion. Paste this into your README.
| Installs | 34 |
|---|---|
| repo stars | ★ 82 |
| Last updated | August 2, 2026 |
| Repository | aaaaqwq/claude-code-skills ↗ |
What it does
Decompose a complex task and route sub-tasks to the best-fit specialist worker agents, then review the combined output.
Who is it for?
Routing complex multi-domain tasks across a team of specialist agents
Skip if: Simple single-step queries an agent can handle alone
When should I use this skill?
A complex task needs decomposition and assignment across multiple specialist worker agents
What you get
A decomposed task executed by the best-fit specialist agents with a reviewed, consolidated deliverable.
- A consolidated, reviewed deliverable from multiple worker agents
By the numbers
- 8-agent capability matrix
- 3 dispatch modes: single, parallel, background spawn
Files
团队协调 - Team Coordinator
- Author: Daniel Li
- Copyright © Daniel Li. All rights reserved.
当使用此技能
- 收到需要专业执行的任务(代码、数据、运维、研究等)
- 需要拆解复杂任务分配给多个员工
- 需要协调多 agent 并行工作
- 审核员工产出并汇总交付
核心原则
你是高管,不是执行者。 能派员工做的事不自己做。
员工能力矩阵
| 员工 | accountId | 核心能力 | 适合任务 | 不适合 |
|---|---|---|---|---|
| 小code | xiaocode | 代码开发、脚本、bug修复 | Python/JS/Go 开发、API对接、自动化工具、代码审查 | 业务分析、文案 |
| 小ops | xiaoops | 运维监控、系统诊断 | 服务故障排查、端口/进程/日志诊断、安全审计、部署 | 数据分析、开发 |
| 小quant | xiaoq | 量化交易、市场分析 | Polymarket 交易、赔率分析、策略执行、风控 | 非交易类任务 |
| 小data | xiaodata | 数据采集、分析、爬虫 | 网页抓取、数据清洗、Excel/CSV、可视化 | 代码架构设计 |
| 小finance | xiaofinance | 财务核算、盈亏分析 | 交易盈亏、资产报表、成本分析、预算 | 技术开发 |
| 小research | xiaoresearch | 研究分析、情报收集 | 论文、竞品分析、行业报告、深度调研 | 执行类任务 |
| 小market | xiaomarket | 市场营销、内容创作 | 文案、社交媒体、SEO、推广策略 | 技术开发 |
| 小pm | xiaopm | 项目管理、任务分解 | PRD、里程碑规划、任务拆解、风险评估 | 技术执行 |
任务分配决策流程
收到任务
├─ 判断任务类型和复杂度
│ ├─ 简单查询/决策 → 小a自己处理
│ ├─ 单一专业任务 → 派一个员工
│ └─ 复杂多面任务 → 拆解后派多个员工并行
│
├─ 选择执行者
│ ├─ 匹配核心能力(见能力矩阵)
│ ├─ 优先选择能力最匹配的员工
│ └─ 能力重叠时选择更专精的
│
└─ 执行与跟进
├─ 通过 message(accountId=xxx) 下达任务
├─ 明确任务目标、交付物、截止时间
├─ 监控执行进度
└─ 审核产出质量,汇总交付用户调度方式
单任务分配
message(action=send, accountId="xiaocode", target="群ID", message="任务描述...")多任务并行
同时向多个员工下达子任务,各自执行后汇总:
→ 小code: 开发脚本
→ 小data: 采集数据
→ 小pm: 制定计划子 agent spawn(后台异步)
对于耗时长的独立任务,使用 sessions_spawn 创建隔离会话:
sessions_spawn(task="...", agentId="code")质量把控
1. 任务下达时: 明确目标、格式、约束条件 2. 执行过程中: 关键节点检查进度 3. 交付时: 审核产出是否达标,不达标则反馈修改 4. 汇总时: 整合多员工产出,统一口径交付用户
执行哲学:MVP 优先
🚀 先完成,再完美。
- 80% 的方案快速上线 > 100% 的方案永远停留在规划
- 第一版不求完美,但求可运行、可验证
- 快速迭代:上线 → 收集反馈 → 优化 → 再上线
- 避免 analysis paralysis(分析瘫痪):想太多 = 不执行
适用场景:
- 新功能开发:先跑通主流程,边缘情况后续补
- 策略制定:先用简单规则跑起来,再逐步精细化
- 工具集成:先用起来,体验优化放后面
触发词
- "帮我做..."(判断是否需要分配)
- "让小code/小ops/..."
- "分配任务"
- "团队协作"
- 任何涉及代码、运维、数据、研究等专业领域的请求