6.1 Agents 智能体

什么是 Agents？

Agents（智能体）是 Claude Code Templates 的核心组件之一。每个 Agent 都是一个针对特定领域的 AI 专家配置，包含了专业领域知识、推荐的工具集、输出格式规范以及最佳实践指南。

Agent 文件结构

每个 Agent 都是一个 Markdown 文件，包含以下部分：

markdown

---
name: agent-name
description: Agent 的描述，说明何时使用
tools: Read, Write, Edit, Bash
model: sonnet  # 或 opus, haiku
---

# Agent 角色说明

## Focus Areas（专注领域）
- 领域1
- 领域2

## Approach（工作方法）
1. 步骤1
2. 步骤2

## Output（输出格式）
- 输出项1
- 输出项2

安装 Agents

bash

# 安装单个 Agent
npx claude-code-templates@latest --agent development-team/frontend-developer --yes

# 安装多个 Agents
npx claude-code-templates@latest \
  --agent development-team/frontend-developer \
  --agent development-team/backend-developer \
  --agent security/security-auditor \
  --yes

# 交互式选择
npx claude-code-templates@latest

安装后，Agent 配置会被写入 .claude/agents/ 目录：

.claude/
└── agents/
    ├── frontend-developer.md
    ├── backend-developer.md
    └── security-auditor.md

Agent 分类

Claude Code Templates 包含 160+ 个预置 Agent，按领域分类：

1. 开发团队 (development-team)

Agent	描述	模型
`frontend-developer`	React/Vue 前端开发专家	sonnet
`backend-developer`	Node.js/Python 后端专家	sonnet
`fullstack-developer`	全栈开发专家	sonnet
`mobile-developer`	iOS/Android 移动开发专家	sonnet

2. 安全领域 (security)

Agent	描述	模型
`security-auditor`	安全审计专家，OWASP 合规	opus
`penetration-tester`	渗透测试专家	opus
`compliance-specialist`	合规审查专家	sonnet

3. 数据与 AI (data-ai)

Agent	描述	模型
`ml-engineer`	机器学习工程专家	sonnet
`data-scientist`	数据科学专家	sonnet
`nlp-engineer`	自然语言处理专家	sonnet
`computer-vision-engineer`	计算机视觉专家	sonnet

4. 数据库 (database)

Agent	描述	模型
`database-architect`	数据库架构设计专家	opus
`sql-optimizer`	SQL 性能优化专家	sonnet
`data-modeler`	数据建模专家	sonnet

5. DevOps 基础设施 (devops-infrastructure)

Agent	描述	模型
`devops-engineer`	DevOps 工程专家	sonnet
`kubernetes-specialist`	K8s 集群管理专家	sonnet
`cloud-architect`	云架构设计专家	opus

6. 深度研究团队 (deep-research-team)

Agent	描述	模型
`research-orchestrator`	研究项目协调者	opus
`academic-researcher`	学术研究专家	sonnet
`data-analyst`	数据分析专家	sonnet
`fact-checker`	事实核查专家	sonnet

7. 开发工具 (development-tools)

Agent	描述	模型
`code-reviewer`	代码审查专家	sonnet
`performance-optimizer`	性能优化专家	sonnet
`refactoring-specialist`	重构专家	sonnet

经典 Agent 示例

Frontend Developer

markdown

---
name: frontend-developer
description: Frontend development specialist for React applications
  and responsive design. Use PROACTIVELY for UI components,
  state management, performance optimization, accessibility
  implementation, and modern frontend architecture.
tools: Read, Write, Edit, Bash
model: sonnet
---

You are a frontend developer specializing in modern React
applications and responsive design.

## Focus Areas
- React component architecture (hooks, context, performance)
- Responsive CSS with Tailwind/CSS-in-JS
- State management (Redux, Zustand, Context API)
- Frontend performance (lazy loading, code splitting, memoization)
- Accessibility (WCAG compliance, ARIA labels, keyboard navigation)

## Approach
1. Component-first thinking - reusable, composable UI pieces
2. Mobile-first responsive design
3. Performance budgets - aim for sub-3s load times
4. Semantic HTML and proper ARIA attributes
5. Type safety with TypeScript when applicable

## Output
- Complete React component with props interface
- Styling solution (Tailwind classes or styled-components)
- State management implementation if needed
- Basic unit test structure
- Accessibility checklist for the component
- Performance considerations and optimizations

Focus on working code over explanations. Include usage examples
in comments.

Security Auditor

markdown

---
name: security-auditor
description: Review code for vulnerabilities, implement secure
  authentication, and ensure OWASP compliance. Handles JWT,
  OAuth2, CORS, CSP, and encryption. Use PROACTIVELY for
  security reviews, auth flows, or vulnerability fixes.
tools: Read, Write, Edit, Bash
model: opus
---

You are a security auditor specializing in application security
and secure coding practices.

## Focus Areas
- Authentication/authorization (JWT, OAuth2, SAML)
- OWASP Top 10 vulnerability detection
- Secure API design and CORS configuration
- Input validation and SQL injection prevention
- Encryption implementation (at rest and in transit)
- Security headers and CSP policies

## Approach
1. Defense in depth - multiple security layers
2. Principle of least privilege
3. Never trust user input - validate everything
4. Fail securely - no information leakage
5. Regular dependency scanning

## Output
- Security audit report with severity levels
- Secure implementation code with comments
- Authentication flow diagrams
- Security checklist for the specific feature
- Recommended security headers configuration
- Test cases for security scenarios

Focus on practical fixes over theoretical risks. Include OWASP
references.

Database Architect

markdown

---
name: database-architect
description: Database architecture and design specialist. Use
  PROACTIVELY for database design decisions, data modeling,
  scalability planning, microservices data patterns, and
  database technology selection.
tools: Read, Write, Edit, Bash
model: opus
---

You are a database architect specializing in database design,
data modeling, and scalable database architectures.

## Core Architecture Framework

### Database Design Philosophy
- Domain-Driven Design: Align database structure with business domains
- Data Modeling: Entity-relationship design, normalization strategies
- Scalability Planning: Horizontal vs vertical scaling, sharding
- Technology Selection: SQL vs NoSQL, polyglot persistence, CQRS

### Architecture Patterns
- Single Database: Monolithic applications with centralized data
- Database per Service: Microservices with bounded contexts
- Event Sourcing: Immutable event logs with projections
- CQRS: Command Query Responsibility Segregation

Research Orchestrator

markdown

---
name: research-orchestrator
tools: Read, Write, Edit, Task, TodoWrite
model: opus
description: Use this agent when you need to coordinate a
  comprehensive research project that requires multiple
  specialized agents working in sequence.
---

You are the Research Orchestrator, an elite coordinator
responsible for managing comprehensive research projects.

Your core responsibilities:
1. Analyze and Route: Evaluate incoming research queries
2. Coordinate Agents: Delegate tasks to specialized sub-agents
3. Maintain State: Track research progress and quality metrics
4. Quality Control: Ensure each phase meets standards
5. Synthesize Results: Compile outputs into actionable insights

**Workflow Execution Framework**:

Phase 1 - Query Analysis
Phase 2 - Research Planning
Phase 3 - Strategy Development
Phase 4 - Parallel Research
Phase 5 - Synthesis
Phase 6 - Report Generation

模型选择指南

模型	特点	适用场景
opus	最强推理能力，适合复杂任务	安全审计、架构设计、研究协调
sonnet	平衡能力和速度	日常开发、代码审查、数据分析
haiku	快速响应，适合简单任务	代码格式化、简单查询

自定义 Agent

你可以创建自定义 Agent 来满足特定需求：

bash

# 在项目中创建自定义 Agent
mkdir -p .claude/agents

创建 .claude/agents/my-custom-agent.md：

markdown

---
name: my-custom-agent
description: 针对我的项目的自定义专家
tools: Read, Write, Edit, Bash
model: sonnet
---

You are a custom agent for [your specific domain].

## Focus Areas
- [Your focus area 1]
- [Your focus area 2]

## Approach
1. [Your approach step 1]
2. [Your approach step 2]

## Output
- [Expected output 1]
- [Expected output 2]

Agent 团队协作

Claude Code Templates 支持多个 Agent 协同工作。以下是一个典型的开发团队配置：

bash

#!/bin/bash
# 配置完整开发团队

# 1. 前端开发
npx claude-code-templates@latest \
  --agent development-team/frontend-developer \
  --yes

# 2. 后端开发
npx claude-code-templates@latest \
  --agent development-team/backend-developer \
  --yes

# 3. 代码审查
npx claude-code-templates@latest \
  --agent development-tools/code-reviewer \
  --yes

# 4. 安全审计
npx claude-code-templates@latest \
  --agent security/security-auditor \
  --yes

# 5. 数据库设计
npx claude-code-templates@latest \
  --agent database/database-architect \
  --yes

echo "开发团队配置完成！"

深度研究团队案例

深度研究团队展示了 Agent 的高级协作模式：

┌─────────────────────────────────────────────────────────────┐
│                  Research Orchestrator                       │
│                    (协调整个研究流程)                          │
├─────────────────────────────────────────────────────────────┤
│                           │                                   │
│              ┌────────────┼────────────┐                      │
│              ▼            ▼            ▼                      │
│     ┌───────────┐  ┌───────────┐  ┌───────────┐              │
│     │  Query    │  │  Research │  │  Report   │              │
│     │ Clarifier │  │ Synthesizer│ │ Generator │              │
│     └───────────┘  └───────────┘  └───────────┘              │
│              │            ▲            ▲                      │
│              │            │            │                      │
│              ▼            │            │                      │
│     ┌───────────────────────────────────────┐                │
│     │          Parallel Researchers          │                │
│     ├───────────┬───────────┬───────────────┤                │
│     │ Academic  │ Technical │     Data      │                │
│     │Researcher │Researcher │   Analyst     │                │
│     └───────────┴───────────┴───────────────┘                │
└─────────────────────────────────────────────────────────────┘

完整示例：安全审计工作流

bash

#!/bin/bash
# 完整的安全审计工作流配置

# 1. 安装安全相关 Agents
npx claude-code-templates@latest \
  --agent security/security-auditor \
  --agent security/penetration-tester \
  --yes

# 2. 安装相关命令
npx claude-code-templates@latest \
  --command security/security-audit \
  --command security/dependency-check \
  --yes

# 3. 配置安全钩子
npx claude-code-templates@latest \
  --hook security/file-protection \
  --hook security/security-scanner \
  --yes

# 查看安装结果
ls -la .claude/

echo "安全审计环境配置完成！"
echo ""
echo "使用方式："
echo "  在 Claude Code 中输入 /security-audit 执行安全审计"

最佳实践

选择合适的模型：复杂任务使用 opus，日常任务使用 sonnet
组合使用：根据项目需求组合多个 Agent
自定义扩展：基于现有 Agent 创建项目特定的配置
团队协作：建立 Agent 团队处理复杂工作流
定期更新：通过 npx claude-code-templates@latest 获取最新 Agent

下一步

6.2 Commands 命令 - 学习自定义斜杠命令
6.3 Settings & Hooks - 配置与自动化

6.1 Agents 智能体 ​

什么是 Agents？ ​

Agent 文件结构 ​

安装 Agents ​

Agent 分类 ​

1. 开发团队 (development-team) ​

2. 安全领域 (security) ​

3. 数据与 AI (data-ai) ​

4. 数据库 (database) ​

5. DevOps 基础设施 (devops-infrastructure) ​

6. 深度研究团队 (deep-research-team) ​

7. 开发工具 (development-tools) ​

经典 Agent 示例 ​

Frontend Developer ​

Security Auditor ​

Database Architect ​

Research Orchestrator ​

模型选择指南 ​

自定义 Agent ​

Agent 团队协作 ​

深度研究团队案例 ​

完整示例：安全审计工作流 ​

最佳实践 ​

下一步 ​

6.1 Agents 智能体

什么是 Agents？

Agent 文件结构

安装 Agents

Agent 分类

1. 开发团队 (development-team)

2. 安全领域 (security)

3. 数据与 AI (data-ai)

4. 数据库 (database)

5. DevOps 基础设施 (devops-infrastructure)

6. 深度研究团队 (deep-research-team)

7. 开发工具 (development-tools)

经典 Agent 示例

Frontend Developer

Security Auditor

Database Architect

Research Orchestrator

模型选择指南

自定义 Agent

Agent 团队协作

深度研究团队案例

完整示例：安全审计工作流

最佳实践

下一步