team-builder
用于组合和派遣并行团队的交互式代理选择器
About this skill
The Team Builder skill provides an AI agent with an interactive menu to dynamically browse and assemble teams of other specialized agents. It's designed for scenarios where a primary AI needs to leverage multiple personas or capabilities from its available agent ecosystem. The skill supports organizing these sub-agents either in a flat directory structure or by domain-specific subdirectories, automatically inferring domains from folder names or file prefixes. Each 'sub-agent' must be defined as a Markdown file containing its complete role prompt (identity, rules, workflow, deliverables), where the first `# Heading` serves as the agent's name and the first paragraph as its description. This empowers the primary agent to intelligently select, combine, and dispatch the right group of agents to tackle complex, multi-faceted tasks that require diverse expertise.
Best use case
Selecting specific agents for a task from a collection of various agent roles, each defined as a Markdown file. Temporarily assembling a cross-domain team, for example, combining a security engineer agent, an SEO specialist, and a software architect agent for a particular project. Browsing and exploring available agents to understand their capabilities before making a selection for a task.
用于组合和派遣并行团队的交互式代理选择器
A dynamically assembled team of specialized AI agents, dispatched by the primary agent, optimized to collaboratively address the specific requirements of a given complex, multi-faceted task.
Practical example
Example input
Agent: "I need to develop a new web feature that requires both backend security review and frontend SEO optimization. Please invoke the Team Builder skill to help me select agents from the 'engineering' and 'marketing' domains. I want to specifically select a 'security-engineer' and an 'seo-specialist'."
Example output
{
"Team Builder Skill invoked. Available domains and agents": {
"engineering": [
"security-engineer",
"software-architect"
],
"marketing": [
"seo-specialist"
],
"sales": [
"discovery-coach"
]
},
"Selected Agents for task 'Develop web feature with security and SEO focus'": [
"security-engineer",
"seo-specialist"
],
"Action": "Team 'security-engineer' and 'seo-specialist' has been assembled and is ready for dispatch."
}When to use this skill
- When the primary AI agent needs to delegate parts of a complex task to specialized sub-agents.
- When the primary agent has access to a library of other agents and needs to dynamically choose which ones to activate.
- When an AI needs to orchestrate a workflow involving multiple distinct expert personas.
- When a task requires a combination of skills that no single agent possesses, necessitating a collaborative team approach.
When not to use this skill
- When the task is simple and can be fully handled by the primary agent alone without needing external specialists.
- When there are no pre-defined sub-agents available for the primary agent to select from.
- When the agent orchestration logic is fixed and doesn't require dynamic selection or browsing of available agents.
- When the structure of sub-agent definitions (Markdown files with specific headings and paragraphs) cannot be met.
Installation
Claude Code / Cursor / Codex
Manual Installation
- Download SKILL.md from GitHub
- Place it in
.claude/skills/team-builder/SKILL.mdinside your project - Restart your AI agent — it will auto-discover the skill
How team-builder Compares
| Feature / Agent | team-builder | Standard Approach |
|---|---|---|
| Platform Support | Claude | Limited / Varies |
| Context Awareness | High | Baseline |
| Installation Complexity | easy | N/A |
Frequently Asked Questions
What does this skill do?
用于组合和派遣并行团队的交互式代理选择器
Which AI agents support this skill?
This skill is designed for Claude.
How difficult is it to install?
The installation complexity is rated as easy. You can find the installation instructions above.
Where can I find the source code?
You can find the source code on GitHub using the link provided at the top of the page.
Related Guides
Best AI Skills for Claude
Explore the best AI skills for Claude and Claude Code across coding, research, workflow automation, documentation, and agent operations.
ChatGPT vs Claude for Agent Skills
Compare ChatGPT and Claude for AI agent skills across coding, writing, research, and reusable workflow execution.
AI Agents for Coding
Browse AI agent skills for coding, debugging, testing, refactoring, code review, and developer workflows across Claude, Cursor, and Codex.
SKILL.md Source
# 团队构建器
用于按需浏览和组合智能体团队的交互式菜单。适用于扁平化或按领域子目录组织的智能体集合。
## 使用场景
* 你拥有多个智能体角色(markdown 文件),并希望为某项任务选择使用哪些智能体
* 你希望从不同领域(例如,安全 + SEO + 架构)临时组建一个团队
* 你希望在决定前先浏览有哪些可用的智能体
## 前提条件
智能体文件必须是包含角色提示(身份、规则、工作流程、交付物)的 markdown 文件。第一个 `# Heading` 用作智能体名称,第一段用作描述。
支持扁平化和子目录两种布局:
**子目录布局** — 领域从文件夹名称推断:
```
agents/
├── engineering/
│ ├── security-engineer.md
│ └── software-architect.md
├── marketing/
│ └── seo-specialist.md
└── sales/
└── discovery-coach.md
```
**扁平化布局** — 领域从共享的文件名前缀推断。当 2 个或更多文件共享同一前缀时,该前缀被视为一个领域。具有唯一前缀的文件归入 "General" 类别。注意:算法在第一个 `-` 处分割,因此多单词领域(例如 `product-management`)应使用子目录布局:
```
agents/
├── engineering-security-engineer.md
├── engineering-software-architect.md
├── marketing-seo-specialist.md
├── marketing-content-strategist.md
├── sales-discovery-coach.md
└── sales-outbound-strategist.md
```
## 配置
智能体目录按顺序探测,结果会被合并:
1. `./agents/**/*.md` + `./agents/*.md` — 项目本地智能体(两种深度)
2. `~/.claude/agents/**/*.md` + `~/.claude/agents/*.md` — 全局智能体(两种深度)
所有位置的结果会合并,并按智能体名称去重。同名情况下,项目本地智能体优先于全局智能体。如果用户指定了自定义路径,则使用该路径代替。
## 工作原理
### 步骤 1:发现可用智能体
使用上述探测顺序在智能体目录中进行全局搜索。排除 README 文件。对于找到的每个文件:
* **子目录布局:** 从父文件夹名称提取领域
* **扁平化布局:** 收集所有文件名前缀(第一个 `-` 之前的文本)。一个前缀只有在出现在 2 个或更多文件名中时才符合领域资格(例如,`engineering-security-engineer.md` 和 `engineering-software-architect.md` 都以 `engineering` 开头 → Engineering 领域)。具有唯一前缀的文件(例如 `code-reviewer.md`, `tdd-guide.md`)归入 "General" 类别
* 从第一个 `# Heading` 提取智能体名称。如果未找到标题,则从文件名派生名称(去除 `.md`,用空格替换连字符,并转换为标题大小写)
* 从标题后的第一段提取一行摘要
如果在探测完所有位置后未找到任何智能体文件,则通知用户:"未找到智能体文件。已检查:\[探测的路径列表]。期望:这些目录中的 markdown 文件。" 然后停止。
### 步骤 2:呈现领域菜单
```
可用的代理领域:
1. 工程领域 — 软件架构师、安全工程师
2. 市场营销 — SEO专家
3. 销售领域 — 发现教练、外拓策略师
请选择领域或指定具体代理(例如:"1,3" 或 "security + seo"):
```
* 跳过智能体数量为零的领域(空目录)
* 显示每个领域的智能体数量
### 步骤 3:处理选择
接受灵活的输入:
* 数字:"1,3" 选择 Engineering 和 Sales 中的所有智能体
* 名称:"security + seo" 对发现的智能体进行模糊匹配
* "all from engineering" 选择该领域中的每个智能体
如果选择的智能体超过 5 个,则按字母顺序列出它们,并要求用户缩小范围:"您选择了 N 个智能体(最多 5 个)。请选择保留哪些,或说 'first 5' 以使用按字母顺序排列的前五个。"
确认选择:
```
选定:安全工程师 + SEO专家
他们应该专注于什么任务?(描述任务)
```
### 步骤 4:并行启动智能体
1. 读取每个所选智能体的 markdown 文件
2. 如果尚未提供,则提示输入任务描述
3. 使用 Agent 工具并行启动所有智能体:
* `subagent_type: "general-purpose"`
* `prompt: "{agent file content}\n\nTask: {task description}"`
* 每个智能体独立运行 — 不需要智能体间通信
4. 如果某个智能体失败(错误、超时或输出为空),则在内联中注明失败(例如,"Security Engineer: failed — \[原因]")并继续处理成功智能体的结果
### 步骤 5:综合结果
收集所有输出并呈现统一的报告:
* 按智能体分组的结果
* 综合部分重点突出:
* 各智能体间的共识
* 建议之间的冲突或矛盾
* 建议的后续步骤
如果只选择了一个智能体,则跳过综合部分,直接呈现输出。
## 规则
* **仅限动态发现。** 切勿硬编码智能体列表。目录中的新文件会自动出现在菜单中。
* **每个团队最多 5 个智能体。** 超过 5 个会产生收益递减和过多的令牌使用。在选择时强制执行。
* **并行分发。** 所有智能体同时运行 — 使用 Agent 工具的并行调用模式。
* **并行 Agent 调用,而非 TeamCreate。** 此技能使用并行 Agent 工具调用来处理独立工作。只有当智能体需要辩论或相互回应时,才需要 TeamCreate(一个用于多智能体对话的 Claude Code 工具)。
## 示例
```
User: team builder
Claude:
可用代理领域:
1. 工程 (2) — 软件架构师,安全工程师
2. 市场营销 (1) — SEO专家
3. 销售 (4) — 探索教练,外拓策略师,提案策略师,销售工程师
4. 支持 (1) — 执行摘要
选择领域或指定具体代理:
User: security + seo
Claude:
已选择:安全工程师 + SEO专家
他们应该处理什么任务?
User: 在发布前审查我的 Next.js 电商网站
[两个代理并行启动,各自将他们的专长应用于代码库]
Claude:
## 安全工程师发现
- [发现内容...]
## SEO专家发现
- [发现内容...]
## 综合分析
双方代理均同意:[...]
冲突点:安全建议的CSP阻止内联样式,SEO需要内联模式标记。解决方案:[...]
后续步骤:[...]
```Related Skills
multi-advisor
Conselho de especialistas — consulta multiplos agentes do ecossistema em paralelo para analise multi-perspectiva de qualquer topico. Ativa personas, especialistas e agentes tecnicos simultaneamente, cada um pela sua otica unica, e consolida em sintese decisoria final.
loki-mode
Version 2.35.0 | PRD to Production | Zero Human Intervention > Research-enhanced: OpenAI SDK, DeepMind, Anthropic, AWS Bedrock, Agent SDK, HN Production (2025)
dispatching-parallel-agents
Use when facing 2+ independent tasks that can be worked on without shared state or sequential dependencies
auto-respawn
Your agent always comes back. Anchor identity and memory on-chain so any new instance can resurrect from just an address — no local state, no single point of failure. Permanent identity and recovery on the Autonomys Network.
klaw Agent Skill
The klaw Agent Skill defines the core environment and capabilities for an AI agent operating within the open-source klaw orchestration platform. It provides tools for file management, web interaction, and dynamic agent creation.
workspace-surface-audit
Audit the active repo, MCP servers, plugins, connectors, env surfaces, and harness setup, then recommend the highest-value ECC-native skills, hooks, agents, and operator workflows. Use when the user wants help setting up Claude Code or understanding what capabilities are actually available in their environment.
ui-demo
Record polished UI demo videos using Playwright. Use when the user asks to create a demo, walkthrough, screen recording, or tutorial video of a web application. Produces WebM videos with visible cursor, natural pacing, and professional feel.
token-budget-advisor
Offers the user an informed choice about how much response depth to consume before answering. Use this skill when the user explicitly wants to control response length, depth, or token budget. TRIGGER when: "token budget", "token count", "token usage", "token limit", "response length", "answer depth", "short version", "brief answer", "detailed answer", "exhaustive answer", "respuesta corta vs larga", "cuántos tokens", "ahorrar tokens", "responde al 50%", "dame la versión corta", "quiero controlar cuánto usas", or clear variants where the user is explicitly asking to control answer size or depth. DO NOT TRIGGER when: user has already specified a level in the current session (maintain it), the request is clearly a one-word answer, or "token" refers to auth/session/payment tokens rather than response size.
skill-comply
Visualize whether skills, rules, and agent definitions are actually followed — auto-generates scenarios at 3 prompt strictness levels, runs agents, classifies behavioral sequences, and reports compliance rates with full tool call timelines
santa-method
Multi-agent adversarial verification with convergence loop. Two independent review agents must both pass before output ships.
safety-guard
Use this skill to prevent destructive operations when working on production systems or running agents autonomously.
repo-scan
Cross-stack source code asset audit — classifies every file, detects embedded third-party libraries, and delivers actionable four-level verdicts per module with interactive HTML reports.