# 开发者输入注释，AI生成代码
# 计算斐波那契数列的前n项
def fibonacci(n):
    # AI建议的代码...
    if n <= 0:
        return []
    elif n == 1:
        return [0]
    
    fibs = [0, 1]
    for i in range(2, n):
        fibs.append(fibs[-1] + fibs[-2])
    return fibs

# 智能体时代的交互模式
$ claude "Create a full-stack task management app with React frontend, 
          FastAPI backend, PostgreSQL database, and Docker deployment. 
          Include user authentication and real-time updates."

[Claude] Planning the implementation...
[Claude] Step 1/12: Creating project structure
[Claude] Step 2/12: Setting up FastAPI with SQLAlchemy models
[Claude] Step 3/12: Implementing JWT authentication
...
[Claude] Step 12/12: Writing deployment documentation

✅ Task completed. Project created at ./task-manager/
📊 Summary: 24 files generated, 3 tests passing, Docker ready

# 智能体任务编排示例
from agent_orchestrator import Task, AgentTeam

# 定义一个复杂的系统开发任务
deploy_task = Task(
    name="deploy_microservice",
    description="Deploy a new user service to production",
    subtasks=[
        Task(name="code_review", 
             agent="senior_reviewer",
             input="Review PR #234 for security issues"),
        Task(name="run_tests",
             agent="qa_engineer",
             input="Run integration tests for user service",
             depends_on=["code_review"]),
        Task(name="build_image",
             agent="devops",
             input="Build Docker image for user service:v2.1.0",
             depends_on=["run_tests"]),
        Task(name="deploy_staging",
             agent="devops",
             input="Deploy to staging environment",
             depends_on=["build_image"]),
        Task(name="smoke_test",
             agent="qa_engineer",
             input="Run smoke tests on staging",
             depends_on=["deploy_staging"]),
        Task(name="deploy_production",
             agent="devops",
             input="Canary deploy to production (10% traffic)",
             depends_on=["smoke_test"])
    ]
)

# 执行任务链
result = AgentTeam.execute(deploy_task)

# 智能体安全约束示例
from dataclasses import dataclass
from typing import List

@dataclass
class SafetyGuardrails:
    """定义智能体的安全约束"""
    
    # 文件系统约束
    allowed_paths: List[str] = None  # 允许访问的目录
    forbidden_patterns: List[str] = None  # 禁止操作的文件模式
    
    # 网络约束
    allowed_domains: List[str] = None  # 允许访问的域名
    max_requests_per_minute: int = 60
    
    # 执行约束
    max_execution_time: int = 300  # 最大执行时间（秒）
    max_file_size: int = 100 * 1024 * 1024  # 最大文件操作大小
    forbidden_commands: List[str] = None  # 禁止执行的系统命令
    
    # 成本约束
    max_api_cost_per_task: float = 10.0  # 最大API调用成本

# 应用约束
guardrails = SafetyGuardrails(
    allowed_paths=["/workspace/project", "/tmp"],
    forbidden_patterns=["*.pem", "*.key", ".env"],
    forbidden_commands=["rm -rf /", "sudo", "mkfs"],
    allowed_domains=["api.github.com", "pypi.org"]
)

# 自动化验证流水线
class AgentOutputValidator:
    def __init__(self):
        self.checks = [
            SyntaxCheck(),
            SecurityCheck(),
            StyleCheck(),
            TestCoverageCheck(),
            PerformanceCheck()
        ]
    
    def validate(self, code_output: str) -> ValidationReport:
        report = ValidationReport()
        
        for check in self.checks:
            result = check.run(code_output)
            report.add_result(result)
            
            if result.severity == "critical":
                report.reject(f"Critical issue found: {result.message}")
                return report
        
        if report.score >= 0.8:
            report.approve()
        else:
            report.request_revision()
        
        return report

# 使用示例
validator = AgentOutputValidator()
report = validator.validate(ai_generated_code)

if report.approved:
    commit_code(report.code)
else:
    send_back_for_revision(report.issues)

# 后端智能体示例：自动API设计
from backend_agent import APIAgent

agent = APIAgent(
    requirements="""
    设计一个电商平台的订单管理API，需要支持：
    - 创建订单（支持优惠券和积分抵扣）
    - 查询订单状态（支持分页和筛选）
    - 取消订单（需要理由和审核）
    - 订单退款（对接支付网关）
    """
)

# 智能体自主完成设计
api_spec = agent.generate_openapi_spec()
database_schema = agent.design_database()
implementation = agent.generate_code(api_spec)
tests = agent.generate_tests(api_spec)

# 输出完整的可运行项目
agent.export_project("./ecommerce-order-api/")