create-team
About
This Claude Skill creates new multi-agent team composition files following the agent-almanac template and registry conventions. It handles team purpose definition, member selection, coordination patterns, task decomposition, and machine-readable configuration. Use it when defining complex multi-agent workflows, review processes, or recurring collaborative task groups in Claude Code.
Quick Install
Claude Code
Recommendednpx skills add pjt222/agent-almanac -a claude-code/plugin add https://github.com/pjt222/agent-almanacgit clone https://github.com/pjt222/agent-almanac.git ~/.claude/skills/create-teamCopy and paste this command in Claude Code to install this skill
Documentation
造團
定調二代以上為含多角、專、階任之多代團。成團檔融於團庫可於 Claude Code 中以名活。
用
- 任需單代不能之多角(如碼評+安審+架評)
- 需含一致角賦與交流之復合作流
- 存代組反復用當正式化
- 複程自分為階或專由異代理
- 欲為 sprint、管、並工定調組
入
- 必:團名(小寫 kebab-case、如
data-pipeline-review) - 必:團目(一段述何問需多代)
- 必:首代(須存於
agents/_registry.yml) - 可:調模(默:hub-and-spoke)之一:
hub-and-spoke、sequential、parallel、timeboxed、adaptive - 可:員數(默:3-4;宜:2-5)
- 可:源材(存流、手冊、臨時團組)
行
一:定團目
明何問需多代同工。有效團目須答:
- 何果團交?(如全評報、部應、sprint 增量)
- 何單代不能? 識至少二異專或角
- 何時當活此團? 定觸件
書目為一段使人或代讀以決活否。
得: 明段述團值、至少二異專識。
敗: 不能識二異專→任或不需團。用含多技之單代。
二:擇首代
首代指揮團。自 agents/_registry.yml 擇代,須:
- 含關團主出之域專
- 可分請為子任予他員
- 可合多評果為一致交
# List all available agents
grep "^ - id:" agents/_registry.yml
首亦須現於團員(首恆為員)。
得: 擇一代為首、確存於代庫。
敗: 無存代合首→先以 create-agent 造(或手 agents/_template.md)。勿造含不存首之團。
三:擇員代
擇 2-5 員(含首)含明、不重責。各員定:
- id:自代庫之代名
- role:短角(如「Quality Reviewer」、「Security Auditor」、「Architecture Reviewer」)
- responsibilities:一句述此員何他員無為
# Verify each candidate agent exists
grep "id: agent-name-here" agents/_registry.yml
驗不重:無二員含同主責。責重→合角或銳界。
得: 擇 2-5 員、各含獨角明責、皆確於代庫。
敗: 需代不存→先造。二員責重→重書以明界或去一員。
四:擇調模
擇最合團流之模。五模與用:
| 模 | 用 | 例團 |
|---|---|---|
| hub-and-spoke | 首分任、集果、合。評審流佳 | r-package-review、gxp-compliance-validation、ml-data-science-review |
| sequential | 各代築於前代出。管與階流佳 | fullstack-web-dev、tending |
| parallel | 諸代同於獨子任。子任無依佳 | devops-platform-engineering |
| timeboxed | 工組為定長迭。含備之持案工佳 | scrum-team |
| adaptive | 團按任自組。未知或高變任佳 | opaque-team |
決導:
- 首須見諸果方出:hub-and-spoke
- 代 B 需代 A 出方始:sequential
- 諸代獨工不見他出:parallel
- 工跨含計式之多迭:timeboxed
- 不能預知任構:adaptive
得: 擇一調模含明因。
敗: 疑→默 hub-and-spoke。最常、多評析流用之。
五:設任分
定典請如何分為員。結為階:
- 設階:首如何析請建任
- 行階:各員作何(按調模並、序、每 sprint)
- 合階:如何集果生終交
各員列典請之 3-5 具任。此任現於「Task Decomposition」文與 CONFIG 塊 tasks 列。
得: 含每員具任之階構分、合擇之調模。
敗: 任泛(如「評事」)→具(如「評碼風合 tidyverse、察試覆、評誤訊質」)。
六:書團檔
複模填諸節:
cp teams/_template.md teams/<team-name>.md
按序填:
- YAML frontmatter:
name、description、lead、version(「1.0.0」)、author、created、updated、tags、coordination、members[](各含 id、role、responsibilities) - 標:
# Team Name(人讀、title case) - 引:一段總結
- 目:此團為何、合何專
- 團組:含 Member、Agent、Role、Focus Areas 列之表
- 調模:文述加流 ASCII 圖
- 任分:含每員具任之階分
- 設:機可讀 CONFIG 塊(見步七)
- 用景:2-3 具景含例用提
- 限:3-5 知限
- See Also:員代檔與相關技/團之聯
得: 全團檔含諸節填、無模留語。
敗: 較存團檔(如 teams/r-package-review.md)驗構。搜模留語如「your-team-name」、「another-agent」尋未填節。
七:書 CONFIG 塊
<!-- CONFIG:START --> 與 <!-- CONFIG:END --> 間 CONFIG 塊予具之機可讀 YAML。構如下:
<!-- CONFIG:START -->
```yaml
team:
name: <team-name>
lead: <lead-agent-id>
coordination: <pattern>
members:
- agent: <agent-id>
role: <role-title>
subagent_type: <agent-id> # Claude Code subagent type for spawning
# ... repeat for each member
tasks:
- name: <task-name>
assignee: <agent-id>
description: <one-line description>
# ... repeat for each task
- name: synthesize-report # final task if hub-and-spoke
assignee: <lead-agent-id>
description: <synthesis description>
blocked_by: [<prior-task-names>] # for dependency ordering
```
<!-- CONFIG:END -->
subagent_type 欄映至 Claude Code 代型。為定於 .claude/agents/ 之代、用代 id 為 subagent_type。用 blocked_by 表任依(如合為諸評任阻)。
得: CONFIG 塊為有效 YAML、諸代合 frontmatter 員列、任依為有效 DAG(無環)。
敗: 驗 YAML 文。驗諸任列之 assignee 合員列之 agent。察 blocked_by 僅引前定任名。
八:加於庫
編 teams/_registry.yml 加新團:
- id: <team-name>
path: <team-name>.md
lead: <lead-agent-id>
members: [<agent-id-1>, <agent-id-2>, ...]
coordination: <pattern>
description: <one-line description matching frontmatter>
更庫首 total_teams 計(現 8、加一團後為 9)。
# Verify the entry was added
grep "id: <team-name>" teams/_registry.yml
得: 新項現於庫、total_teams 計增一。
敗: 團名已存於庫→擇異名或更存項。驗 YAML 縮合存項。
九:行 README 自動
自更庫重生 README:
npm run update-readmes
此更 teams/README.md 與任引團數之含 <!-- AUTO:START --> / <!-- AUTO:END --> 標之他檔動節。
得: 令出零、teams/README.md 列新團。
敗: 行 npm run check-readmes 見何檔未同。本敗→驗 package.json 於庫根、js-yaml 裝(npm install)。
十:驗團活
試團可於 Claude Code 中活:
User: Use the <team-name> team to <typical task description>
Claude 讀 teams/<team-name>.md、取 CONFIG 塊、調活:
- 以團名與述調
TeamCreate - 以
Agent具按各員 CONFIG 塊之subagent_type生員 - 以 CONFIG 塊之
blocked_by依調TaskCreate - 首代按調模調工
注:團非自發現於 .claude/teams/。求時 Claude 自 teams/ 直讀定。
得: Claude 讀團檔、以 TeamCreate 建團、生正代、循調模。
敗: 驗團檔於 teams/<team-name>.md(非子目)。察諸員代存於 agents/。確 CONFIG 塊含各員 subagent_type 之有效 YAML。確團列於 teams/_registry.yml。
十一:架譯
諸團必。此步施於人作者與循此程之 AI 代。勿略——缺譯積為陳備。
承新團後即為諸 4 locales 架譯檔:
for locale in de zh-CN ja es; do
npm run translate:scaffold -- teams <team-name> "$locale"
done
續譯各檔之架詞(碼塊與 ID 留英)。終重生態檔:
npm run translation:status
得: i18n/{de,zh-CN,ja,es}/teams/<team-name>.md 建四檔、source_commit 皆合現 HEAD。npm run validate:translations 顯零陳警於新團。
敗: 架敗→驗團存於 teams/_registry.yml。態檔不更→顯行 npm run translation:status。
驗
- 團檔存於
teams/<team-name>.md - YAML frontmatter 無誤解析
- 諸必欄存:
name、description、lead、version、author、coordination、members[] - 各員含
id、role、responsibilities - 諸節存:Purpose、Team Composition、Coordination Pattern、Task Decomposition、Configuration、Usage Scenarios、Limitations、See Also
- CONFIG 塊存於
<!-- CONFIG:START -->與<!-- CONFIG:END -->標間 - CONFIG 塊 YAML 有效可解
- 諸員代 id 存於
agents/_registry.yml - 首代現於員列
- 無二員同主責
- 團列於
teams/_registry.yml含正路、首、員、調 - 庫中
total_teams計增 -
npm run update-readmes無誤畢
忌
- 員過多:過 5 員之團難調。分任與合果之過負勝加角之益。分二團或減至要專
- 責重:二員皆「評碼質」→見衝而首費時除重。各員須含明異焦
- 誤調模:代需他出時用 hub-and-spoke(當 sequential)、代可獨工時用 sequential(當 parallel)。察步四決導
- 缺 CONFIG 塊:CONFIG 塊非可文飾。為 Claude 調
TeamCreate、代生、任建之機可讀規。無之→團僅可以臨文解活、不靠 - 首代非員:首須現為員含己角責。僅「調」無實工之首費位。予首具評或合責
參
create-skill- 循同元模造 SKILL.md 檔create-agent- 造為團員之代定commit-changes- 承新團檔與庫更
GitHub Repository
Related Skills
content-collections
MetaThis skill provides a production-tested setup for Content Collections, a TypeScript-first tool that transforms Markdown/MDX files into type-safe data collections with Zod validation. Use it when building blogs, documentation sites, or content-heavy Vite + React applications to ensure type safety and automatic content validation. It covers everything from Vite plugin configuration and MDX compilation to deployment optimization and schema validation.
polymarket
MetaThis skill enables developers to build applications with the Polymarket prediction markets platform, including API integration for trading and market data. It also provides real-time data streaming via WebSocket to monitor live trades and market activity. Use it for implementing trading strategies or creating tools that process live market updates.
creating-opencode-plugins
MetaThis skill helps developers create OpenCode plugins that hook into 25+ event types like commands, files, and LSP operations. It provides the plugin structure, event API specifications, and implementation patterns for JavaScript/TypeScript modules. Use it when you need to intercept, monitor, or extend the OpenCode AI assistant's lifecycle with custom event-driven logic.
sglang
MetaSGLang is a high-performance LLM serving framework that specializes in fast, structured generation for JSON, regex, and agentic workflows using its RadixAttention prefix caching. It delivers significantly faster inference, especially for tasks with repeated prefixes, making it ideal for complex, structured outputs and multi-turn conversations. Choose SGLang over alternatives like vLLM when you need constrained decoding or are building applications with extensive prefix sharing.
