Kimi Claw 深度部署与全场景实战指南

version: '3.8'
services:
  kimi-byoc-claw:
    image: openclaw/openclaw:latest # 官方最新镜像 (Official latest image)
    container_name: kimi_byoc_instance
    restart: always # 意外崩溃自动重启 (Auto-restart on crash)
    ports:
      - "3000:3000"
    environment:
      # 这里的 Token 从 kimi.com 的 BYOC 配置页获取
      # Get this token from the BYOC settings page on kimi.com
      - BYOC_CONNECTION_TOKEN=sk-byoc-dummy-token-xxxxxx
      - KIMI_API_KEY=sk-dummy-api-key-xxxxxx
      - TZ=Asia/Shanghai # 强制时区，防止定时任务错乱 (Force timezone)
    volumes:
      # 映射本地目录用于存储持久化数据和记忆
      # Map local directory for persistent data and memory
      - ./claw_data:/app/data
      # 映射日志目录以便日后排错
      # Map log directory for troubleshooting
      - ./logs:/app/logs

docker-compose up -d

// pages/index/index.js
Page({
  data: {
    basePrice: '',      // 基础单价 (Base price per unit)
    quantity: '',       // 采购数量 (Purchase quantity)
    // 界面显示的代理级别文案 (Display text for agent levels)
    levels: ['省级总代 (0.5折)', '市级代理 (0.7折)', '普通分销商 (0.9折)'], 
    levelIndex: 0,      // 当前用户选择的下拉框索引 (Current selected index)
    // 对应的底层真实计算折扣率 (Actual discount rates for math)
    discountRates: [0.5, 0.7, 0.9], 
    originalTotal: 0,   // 折前总计总价 (Total before discount)
    discount: 1,        // 当前应用的真实折扣 (Current applied discount)
    finalPrice: 0       // 客户最终需要支付的金额 (Final price to pay)
  },

  // 监听单价输入框的内容变化 (Listen to base price input changes)
  onBasePriceInput(e) {
    this.setData({
      basePrice: e.detail.value
    });
  },

  // 监听数量输入框的内容变化 (Listen to quantity input changes)
  onQuantityInput(e) {
    this.setData({
      quantity: e.detail.value
    });
  },

  // 监听下拉选择器的切换事件 (Listen to picker changes)
  onLevelChange(e) {
    this.setData({
      levelIndex: e.detail.value
    });
  },

  // 核心功能：执行价格计算 (Core function: Execute price calculation)
  calculatePrice() {
    // 将字符串转换为浮点数和整数 (Parse strings to float/int)
    const price = parseFloat(this.data.basePrice);
    const qty = parseInt(this.data.quantity, 10);
    
    // 表单基础校验，防止恶意输入或空值提交 (Basic validation)
    if (isNaN(price) || price <= 0) {
      wx.showToast({ title: '请输入正确的商品单价', icon: 'none' });
      return;
    }
    if (isNaN(qty) || qty <= 0) {
      wx.showToast({ title: '请输入正确的采购数量', icon: 'none' });
      return;
    }

    // 根据用户选择的索引，获取对应的真实折扣率 (Get the actual discount rate)
    const currentDiscount = this.data.discountRates[this.data.levelIndex];
    
    // 关键步骤：计算折前总价
    // 为了避免 JavaScript 经典的 0.1+0.2=0.30000000000000004 精度丢失问题，
    // 我们先将金额放大 100 倍转化为整数计算，然后再缩小 100 倍。
    const original = Math.round(price * 100) * qty / 100;
    
    // 计算最终总价，并强制保留两位小数 (Calculate final price and fix to 2 decimal places)
    const final = (original * currentDiscount).toFixed(2);

    // 将计算结果更新到页面数据中，触发视图渲染 (Update page data to render view)
    this.setData({
      originalTotal: original.toFixed(2),
      discount: currentDiscount,
      finalPrice: final
    });
    
    // 调用微信原生的震动反馈 API，提升用户体验 (Trigger haptic feedback)
    wx.vibrateShort();
  }
})

"""
=============================================================================
脚本名称: 科技公司销售部开年工作计划 PPT 自动化生成器
执行环境: Python 3.10++
核心依赖: pip install python-pptx
功能描述: 基于 python-pptx 库，动态生成包含特定业务数据的商业汇报演示文稿。
=============================================================================
"""

import os
# 导入 PPT 核心处理对象 (Import core presentation object)
from pptx import Presentation
# 导入尺寸和单位转换工具 (Import measurement utilities)
from pptx.util import Inches, Pt
# 导入颜色处理工具 (Import color utilities)
from pptx.dml.color import RGBColor

def generate_sales_kickoff_ppt(output_filename="2026_Sales_Plan.pptx"):
    """
    主函数：按预定结构生成并保存 PPT 文件
    """
    print(">>> 正在初始化 PPT 实例对象...")
    # 创建一个空白幻灯片对象 (Create a blank presentation object)
    prs = Presentation()
    
    # 预先获取内置的幻灯片母版布局索引 (Get built-in slide layout indices)
    # 通常 0 是标题布局，1 是标题+带项目符号的内容布局
    layout_title = prs.slide_layouts[0]
    layout_bullet = prs.slide_layouts[1]

    # -----------------------------------------------------------------------
    # 第 1 页：生成封面 (Slide 1: Cover)
    # -----------------------------------------------------------------------
    print(">>> 正在渲染第 1 页：封面图...")
    slide1 = prs.slides.add_slide(layout_title)
    
    # 获取标题和副标题的占位符形状 (Get title and subtitle placeholders)
    title_shape = slide1.shapes.title
    subtitle_shape = slide1.placeholders[1]

    # 注入文本内容 (Inject text content)
    title_shape.text = "破局·聚力·共赢\n销售部 2026 开年计划"
    subtitle_shape.text = "汇报人：销售总监\n生成日期：2026年2月"

    # 遍历标题中的段落，强制设置字体和颜色，增加商业高级感
    for paragraph in title_shape.text_frame.paragraphs:
        paragraph.font.name = '微软雅黑'
        paragraph.font.size = Pt(44)
        paragraph.font.bold = True
        paragraph.font.color.rgb = RGBColor(0, 82, 217) # 设定为沉稳的科技蓝

    # -----------------------------------------------------------------------
    # 第 2 页：核心业绩回顾与痛点 (Slide 2: Review and Pain Points)
    # -----------------------------------------------------------------------
    print(">>> 正在渲染第 2 页：业绩回顾...")
    slide2 = prs.slides.add_slide(layout_bullet)
    slide2.shapes.title.text = "业绩回顾与痛点剖析"
    
    # 获取正文文本框对象 (Get body text frame)
    tf2 = slide2.shapes.placeholders[1].text_frame
    tf2.text = "去年度核心业务指标总结："
    
    # 动态添加带有层级的子弹点 (Add hierarchical bullet points)
    p = tf2.add_paragraph()
    p.text = "✅ 亮点：大客户直销业务同比增长 24%，超额完成全年目标。"
    p.level = 1
    p.font.name = '微软雅黑'
    
    p = tf2.add_paragraph()
    p.text = "❌ 痛点一：中小企业 SaaS 订阅版块续费率偏低，仅维持在 68%。"
    p.level = 1
    p.font.name = '微软雅黑'

    p = tf2.add_paragraph()
    p.text = "❌ 痛点二：跨部门协同审批流程繁琐，导致线索转化周期被拉长。"
    p.level = 1
    p.font.name = '微软雅黑'

    # -----------------------------------------------------------------------
    # 第 3 页：新年三大战役 (Slide 3: Strategic Battles)
    # -----------------------------------------------------------------------
    print(">>> 正在渲染第 3 页：战略目标...")
    slide3 = prs.slides.add_slide(layout_bullet)
    slide3.shapes.title.text = "新年三大战役规划"
    
    tf3 = slide3.shapes.placeholders[1].text_frame
    tf3.text = "为实现年度业绩翻倍，确立以下核心战略："
    
    # 战役一
    p = tf3.add_paragraph()
    p.text = "战役一：行业大客户攻坚战"
    p.level = 1
    p.font.bold = True
    p = tf3.add_paragraph()
    p.text = "深度锁定医疗与政务赛道，定向培育 5 个千万级标杆案例项目。"
    p.level = 2

    # 战役二
    p = tf3.add_paragraph()
    p.text = "战役二：生态渠道裂变战"
    p.level = 1
    p.font.bold = True
    p = tf3.add_paragraph()
    p.text = "招募并赋能核心城市合伙人，完善返佣机制，激活下沉市场。"
    p.level = 2
    
    # 战役三
    p = tf3.add_paragraph()
    p.text = "战役三：内部效能提升战"
    p.level = 1
    p.font.bold = True
    p = tf3.add_paragraph()
    p.text = "全面引入 AI 工具（如专属智能报价助手原型）赋能一线销售。"
    p.level = 2

    # -----------------------------------------------------------------------
    # 第 4 页：Q1 冲刺行动方案 (Slide 4: Q1 Action Plan)
    # -----------------------------------------------------------------------
    print(">>> 正在渲染第 4 页：行动时间表...")
    slide4 = prs.slides.add_slide(layout_bullet)
    slide4.shapes.title.text = "Q1 关键路径与行动时间轴"
    
    tf4 = slide4.shapes.placeholders[1].text_frame
    # 清空母版默认附带的格式，准备写入纯文本排版
    tf4.clear() 
    
    actions = [
        "【2月上旬】完善产品报价体系，内测全新的代理商分级计算系统。",
        "【2月中旬】举办全国渠道招商线上大会，定向释放区域独家代理权。",
        "【3月初】组织销售全员进行新版本产品知识竞标赛与实战话术演练。",
        "【3月底】针对去年沉睡线索库进行第一轮集中促活与公海捞鱼行动。"
    ]
    
    # 循环写入文本并统一样式
    for action in actions:
        p = tf4.add_paragraph()
        p.text = action
        p.font.size = Pt(22)
        p.font.name = '微软雅黑'
        p.space_after = Pt(18) # 设置段后间距，避免文字挤在一起

    # -----------------------------------------------------------------------
    # 文件保存逻辑 (File Saving Logic)
    # -----------------------------------------------------------------------
    try:
        prs.save(output_filename)
        # 获取绝对路径，方便用户查找
        full_path = os.path.abspath(output_filename)
        print(f"\n>>> 成功！PPT 演示文稿已生成并保存在：\n>>> {full_path}")
    except PermissionError:
        print(f"\n[错误] 权限拒绝：文件 {output_filename} 被占用或无写入权限。请确保该文件当前未被其他程序打开。")
    except Exception as e:
        print(f"\n[错误] 保存文件时发生未知异常: {str(e)}")

# 脚本主入口
if __name__ == "__main__":
    generate_sales_kickoff_ppt()

pip install httpx aiofiles

"""
Kimi Claw 主动触发探针 (Active Trigger Probe for Kimi Claw)
环境要求 (Requirements): Python 3.10++
依赖 (Dependencies): httpx, aiofiles
"""

import asyncio
import httpx
import logging
from pathlib import Path
from datetime import datetime
import json

# 配置日志格式，输出到终端 (Configure logging format, output to console)
logging.basicConfig(
    level=logging.INFO,
    format='%(asctime)s - %(levelname)s - %(message)s'
)

# Kimi Claw 提供的专属触发 Webhook 地址 (Replace with your actual BYOC or cloud webhook)
KIMI_WEBHOOK_URL = "https://api.kimi.com/claw/v1/webhook/trigger/dummy-endpoint"
# 鉴权 Token (Authentication token)
AUTHORIZATION_TOKEN = "Bearer sk-dummy-webhook-token"

async def trigger_kimi_agent(error_snippet: str, context_info: str) -> None:
    """
    通过 HTTPX 异步向 Kimi Claw 发送分析请求。
    Asynchronously send analysis request to Kimi Claw via HTTPX.
    """
    headers = {
        "Authorization": AUTHORIZATION_TOKEN,
        "Content-Type": "application/json"
    }
    
    # 构造给 AI 的 prompt 负载 (Construct prompt payload for AI)
    payload = {
        "event_type": "server_error_alert",
        "instruction": (
            "你是一个高级运维专家。请分析以下错误日志，并给出排查思路和修复建议。"
            "分析完成后，将结果推送到我的飞书频道。"
        ),
        "data": {
            "error_log": error_snippet,
            "server_context": context_info,
            "timestamp": datetime.now().isoformat()
        }
    }
    
    async with httpx.AsyncClient(timeout=30.0) as client:
        try:
            logging.info("正在唤醒 Kimi Claw Agent 进行分析... (Waking up Kimi Claw Agent...)")
            response = await client.post(KIMI_WEBHOOK_URL, headers=headers, json=payload)
            response.raise_for_status() # 检查 HTTP 状态码 (Check HTTP status code)
            logging.info(f"触发成功！(Trigger successful!) 响应: {response.status_code}")
        except httpx.HTTPStatusError as exc:
            logging.error(f"触发失败，HTTP 错误 (HTTP error occurred): {exc.response.status_code} - {exc.response.text}")
        except Exception as e:
            logging.error(f"网络请求发生未知异常 (Unknown network exception): {str(e)}")

async def monitor_log_file(filepath: str, keyword: str = "ERROR"):
    """
    异步监控日志文件末尾，类似 tail -f 功能。
    Asynchronously tail the log file for specific keywords.
    """
    log_path = Path(filepath)
    if not log_path.exists():
        logging.error(f"日志文件不存在 (Log file does not exist): {filepath}")
        return

    logging.info(f"开始监控日志 (Started monitoring log): {filepath}")
    
    # 使用 aiofiles 异步读取 (Use aiofiles for async reading)
    import aiofiles
    async with aiofiles.open(filepath, mode='r', encoding='utf-8') as f:
        # 跳到文件末尾 (Seek to the end of the file)
        await f.seek(0, 2)
        while True:
            line = await f.readline()
            if not line:
                await asyncio.sleep(1) # 如果没有新内容，休眠 1 秒 (Sleep if no new content)
                continue
            
            if keyword in line:
                logging.warning(f"检测到异常日志 (Detected abnormal log): {line.strip()}")
                # 收集上下文信息 (Collect context info)
                context = f"监控节点: 核心数据库服务器 (Node: Core DB Server), 触发词: {keyword}"
                # 触发 Agent (Trigger Agent)
                await trigger_kimi_agent(line.strip(), context)

if __name__ == "__main__":
    # 假设监控宝塔环境下的某 Nginx 或 Python 错误日志
    # Assuming we are monitoring an Nginx or Python error log
    target_log = "/var/log/syslog" 
    
    try:
        # Python 3.10+ 推荐的异步入口点 (Recommended async entry point in Python 3.10+)
        asyncio.run(monitor_log_file(target_log, "CRITICAL_ERROR"))
    except KeyboardInterrupt:
        logging.info("监控已手动停止。(Monitoring stopped manually.)")

[Unit]
Description=Kimi Claw Log Monitor Probe
After=network.target

[Service]
Type=simple
User=root
# 请修改为你的实际 Python 路径和脚本路径
# Modify to your actual Python path and script path
ExecStart=/usr/bin/python3.14 /opt/scripts/log_watcher.py
Restart=on-failure
RestartSec=5s

[Install]
WantedBy=multi-user.target