### Install and Configure Xianyu AutoAgent

Source: https://github.com/shaxiu/xianyuautoagent/blob/main/README.md

Commands to clone the repository, install dependencies, and set up the required environment variables.

```bash
1. 克隆仓库
git clone https://github.com/shaxiu/XianyuAutoAgent.git
cd XianyuAutoAgent

2. 安装依赖
pip install -r requirements.txt

3. 配置环境变量
创建一个 `.env` 文件，包含以下内容，也可直接重命名 `.env.example` ：
#必配配置
API_KEY=apikey通过模型平台获取
COOKIES_STR=填写网页端获取的cookie
MODEL_BASE_URL=模型地址
MODEL_NAME=模型名称
#可选配置
TOGGLE_KEYWORDS=接管模式切换关键词，默认为句号（输入句号切换为人工接管，再次输入则切换AI接管）
SIMULATE_HUMAN_TYPING=True/False #模拟人工回复延迟

注意：默认使用的模型是通义千问，如需使用其他API，请自行修改.env文件中的模型地址和模型名称；
COOKIES_STR自行在闲鱼网页端获取cookies(网页端F12打开控制台，选择Network，点击Fetch/XHR,点击一个请求，查看cookies)

4. 创建提示词文件prompts/*_prompt.txt（也可以直接将模板名称中的_example去掉），否则默认读取四个提示词模板中的内容
```

--------------------------------

### Run the Main Program

Source: https://github.com/shaxiu/xianyuautoagent/blob/main/README.md

Command to start the Xianyu AutoAgent service.

```bash
python main.py
```

--------------------------------

### Configure Environment Variables

Source: https://context7.com/shaxiu/xianyuautoagent/llms.txt

Define system configuration including API keys, model settings, and heartbeat intervals in a .env file.

```bash
# .env 配置文件示例

# 必需配置
API_KEY=sk-xxxxxxxxxxxxx                    # 通义千问API密钥（百炼平台获取）
COOKIES_STR=unb=123;cookie2=xxx;...         # 闲鱼登录Cookie

# 模型配置
MODEL_BASE_URL=https://dashscope.aliyuncs.com/compatible-mode/v1
MODEL_NAME=qwen-max                         # 可选: qwen-plus, qwen-turbo

# 心跳配置
HEARTBEAT_INTERVAL=15                       # 心跳间隔（秒）
HEARTBEAT_TIMEOUT=5                         # 心跳超时（秒）

# Token刷新配置
TOKEN_REFRESH_INTERVAL=3600                 # Token刷新间隔（秒，默认1小时）
TOKEN_RETRY_INTERVAL=300                    # Token重试间隔（秒，默认5分钟）

# 人工接管配置
TOGGLE_KEYWORDS=。                          # 切换人工/自动模式的关键词
MANUAL_MODE_TIMEOUT=3600                    # 人工模式超时时间（秒）
MESSAGE_EXPIRE_TIME=300000                  # 消息过期时间（毫秒，默认5分钟）

# 其他配置
LOG_LEVEL=INFO                              # 日志级别: DEBUG, INFO, WARNING, ERROR
SIMULATE_HUMAN_TYPING=False                 # 是否模拟人工输入延迟
```

--------------------------------

### Initialize XianyuLive WebSocket Listener

Source: https://context7.com/shaxiu/xianyuautoagent/llms.txt

The main entry point for establishing WebSocket connections, managing heartbeats, and handling token refreshes.

```python
import asyncio
import os
from dotenv import load_dotenv
from XianyuAgent import XianyuReplyBot

# 加载环境变量
load_dotenv()

class XianyuLive:
    def __init__(self, cookies_str):
        self.xianyu = XianyuApis()
        self.base_url = 'wss://wss-goofish.dingtalk.com/'
        self.cookies_str = cookies_str
        self.cookies = trans_cookies(cookies_str)
        self.xianyu.session.cookies.update(self.cookies)
        self.myid = self.cookies['unb']
        self.device_id = generate_device_id(self.myid)
        self.context_manager = ChatContextManager()

        # 心跳配置
        self.heartbeat_interval = int(os.getenv("HEARTBEAT_INTERVAL", "15"))
        self.heartbeat_timeout = int(os.getenv("HEARTBEAT_TIMEOUT", "5"))

        # Token刷新配置
        self.token_refresh_interval = int(os.getenv("TOKEN_REFRESH_INTERVAL", "3600"))

        # 人工接管配置
        self.manual_mode_timeout = int(os.getenv("MANUAL_MODE_TIMEOUT", "3600"))
        self.toggle_keywords = os.getenv("TOGGLE_KEYWORDS", "。")

# 启动机器人
cookies_str = os.getenv("COOKIES_STR")
bot = XianyuReplyBot()
xianyu_live = XianyuLive(cookies_str)
asyncio.run(xianyu_live.main())
```

--------------------------------

### Deploy with Docker

Source: https://context7.com/shaxiu/xianyuautoagent/llms.txt

Use Docker Compose to manage the service container and persistent volumes.

```yaml
# docker-compose.yml
version: '3.8'
services:
  xianyu-agent:
    build: .
    container_name: xianyu-autoagent
    restart: unless-stopped
    volumes:
      - ./data:/app/data          # 持久化数据库
      - ./.env:/app/.env          # 环境变量配置
      - ./prompts:/app/prompts    # 自定义提示词
    environment:
      - TZ=Asia/Shanghai
```

```bash
# 启动服务
docker-compose up -d

# 查看日志
docker-compose logs -f

# 停止服务
docker-compose down
```

--------------------------------

### Provide Technical Support with TechAgent

Source: https://context7.com/shaxiu/xianyuautoagent/llms.txt

Handles product specifications and comparisons, supporting web search integration for up-to-date information.

```python
from XianyuAgent import TechAgent

tech_prompt = """
你是一位专业的产品技术顾问，擅长回答产品规格、参数、适配性等技术问题。
回答要简洁专业，突出产品优势。
"""

tech_agent = TechAgent(client, tech_prompt, safety_filter)

# 技术咨询示例 - 启用联网搜索
reply = tech_agent.generate(
    user_msg="这个耳机的降噪效果怎么样，和Bose QC45比呢",
    item_desc='{"title": "索尼WH-1000XM5", "desc": "头戴式主动降噪耳机"}',
    context=""
)
# TechAgent 会通过 enable_search=True 联网获取最新对比信息
print(reply)
# 输出示例: "XM5降噪业界顶级，比QC45低频降噪更强，佩戴也更舒适"
```

--------------------------------

### Utilize Xianyu Utility Functions

Source: https://context7.com/shaxiu/xianyuautoagent/llms.txt

Provides helper functions for cookie parsing, ID generation, API signing, and message decryption.

```python
from utils.xianyu_utils import (
    trans_cookies,
    generate_mid,
    generate_uuid,
    generate_device_id,
    generate_sign,
    decrypt
)

# 解析Cookie字符串为字典
cookie_str = "unb=123456; cookie2=abcdef; _m_h5_tk=token_12345"
cookies = trans_cookies(cookie_str)
print(cookies)
# {'unb': '123456', 'cookie2': 'abcdef', '_m_h5_tk': 'token_12345'}

# 生成消息ID（用于WebSocket通信）
mid = generate_mid()
print(mid)  # "1234561234567890123 0"

# 生成UUID（用于消息唯一标识）
uuid = generate_uuid()
print(uuid)  # "-16234567890121"

# 生成设备ID（基于用户ID）
device_id = generate_device_id("123456")
print(device_id)  # "a1b2c3d4-e5f6-4789-abcd-ef0123456789-123456"

# 生成API签名
timestamp = "1234567890123"
token = "abcdef123456"
data = '{"itemId":"123456"}'
sign = generate_sign(timestamp, token, data)
print(sign)  # "e10adc3949ba59abbe56e057f20f883e" (MD5签名)

# 解密WebSocket消息（MessagePack格式）
encrypted_data = "gqExo..."  # Base64编码的MessagePack数据
decrypted = decrypt(encrypted_data)
print(decrypted)  # JSON格式的消息内容
```

--------------------------------

### Generate AI Replies with XianyuReplyBot

Source: https://context7.com/shaxiu/xianyuautoagent/llms.txt

Core class for initializing agents, loading prompts, and generating context-aware responses.

```python
from XianyuAgent import XianyuReplyBot

# 初始化机器人（自动从环境变量读取API配置）
bot = XianyuReplyBot()

# 商品信息
item_desc = '''当前商品的信息如下：{
    "title": "索尼WH-1000XM5头戴式降噪耳机",
    "desc": "99新，仅拆封试听，支持LDAC",
    "price_range": "¥1800",
    "total_stock": 1,
    "sku_details": [{"spec": "黑色", "price": 1800.0, "stock": 1}]
}'''

# 对话历史上下文
context = [
    {"role": "user", "content": "你好，这个耳机还在吗"},
    {"role": "assistant", "content": "在的亲，随时可拍"}
]

# 生成回复
user_message = "能便宜点吗，1500行不行"
reply = bot.generate_reply(user_message, item_desc, context)
print(f"机器人回复: {reply}")
# 输出示例: "亲，这个价格已经很实惠了，诚心要可以小刀"

# 检查最后识别的意图
print(f"识别意图: {bot.last_intent}")  # 输出: price

# 重新加载提示词（热更新）
bot.reload_prompts()
```

--------------------------------

### Configure IntentRouter for Message Classification

Source: https://context7.com/shaxiu/xianyuautoagent/llms.txt

Implements a three-level routing strategy using keywords, regex patterns, and LLM fallback.

```python
from XianyuAgent import IntentRouter, ClassifyAgent
from openai import OpenAI
import os

# 创建分类Agent
client = OpenAI(
    api_key=os.getenv("API_KEY"),
    base_url=os.getenv("MODEL_BASE_URL", "https://dashscope.aliyuncs.com/compatible-mode/v1")
)

# 意图路由器的规则配置
router_rules = {
    'tech': {
        'keywords': ['参数', '规格', '型号', '连接', '对比'],
        'patterns': [r'和.+比']
    },
    'price': {
        'keywords': ['便宜', '价', '砍价', '少点'],
        'patterns': [r'\d+元', r'能少\d+']
    }
}

# 意图检测示例
user_messages = [
    "这个耳机支持什么蓝牙版本",  # -> tech (技术类关键词)
    "能便宜200吗",              # -> price (价格类关键词 + 正则)
    "你好在吗",                  # -> default (大模型兜底)
    "和AirPods Pro比怎么样",    # -> tech (正则匹配)
]

for msg in user_messages:
    # 路由器会自动判断意图
    intent = router.detect(msg, item_desc, context_str)
    print(f"消息: {msg} -> 意图: {intent}")
```

--------------------------------

### Handle Bargaining with PriceAgent

Source: https://context7.com/shaxiu/xianyuautoagent/llms.txt

Specialized agent for price negotiations using dynamic temperature settings and safety filters.

```python
from XianyuAgent import PriceAgent
from openai import OpenAI

client = OpenAI(
    api_key=os.getenv("API_KEY"),
    base_url="https://dashscope.aliyuncs.com/compatible-mode/v1"
)

# 议价提示词
price_prompt = """
【核心策略】
1. 优惠上限：设定明确的优惠上限（如100元或商品价格的10%）
2. 梯度让步：根据议价次数逐步增加优惠幅度
3. 价值强调：突出产品品质和价值
"""

def safety_filter(text):
    blocked = ["微信", "QQ", "支付宝", "银行卡", "线下"]
    return "[安全提醒]请通过平台沟通" if any(p in text for p in blocked) else text

price_agent = PriceAgent(client, price_prompt, safety_filter)

# 根据议价次数动态调整温度
# 第1次议价: temperature = 0.3 + 1*0.15 = 0.45 (较保守)
# 第3次议价: temperature = 0.3 + 3*0.15 = 0.75 (更灵活)
# 第5次议价: temperature = min(0.3 + 5*0.15, 0.9) = 0.9 (最大值)

reply = price_agent.generate(
    user_msg="1600能出吗",
    item_desc='{"title": "耳机", "price_range": "¥1800"}',
    context="user: 能便宜吗\nassistant: 诚心要可以小刀",
    bargain_count=2  # 第二次议价
)
print(reply)  # 输出示例: "1700最低了，真的没利润了"
```

--------------------------------

### Interact with Xianyu APIs

Source: https://context7.com/shaxiu/xianyuautoagent/llms.txt

XianyuApis handles authentication, token retrieval, and item information queries with automatic signature and cookie management.

```python
from XianyuApis import XianyuApis
from utils.xianyu_utils import trans_cookies, generate_device_id

# 初始化API客户端
api = XianyuApis()

# 设置Cookie
cookies_str = "unb=123456; cookie2=xxxxx; _m_h5_tk=token_xxxxx; ..."
cookies = trans_cookies(cookies_str)
api.session.cookies.update(cookies)

# 检查登录状态
is_logged_in = api.hasLogin()
if is_logged_in:
    print("登录状态有效")

# 获取WebSocket连接Token
user_id = cookies['unb']
device_id = generate_device_id(user_id)
token_result = api.get_token(device_id)

if 'data' in token_result:
    access_token = token_result['data']['accessToken']
    print(f"Token获取成功: {access_token[:20]}...")

# 获取商品详情
item_id = "123456789"
item_result = api.get_item_info(item_id)

if 'data' in item_result and 'itemDO' in item_result['data']:
    item = item_result['data']['itemDO']
    print(f"商品标题: {item.get('title')}")
    print(f"商品价格: {item.get('soldPrice') / 100}元")
    print(f"商品描述: {item.get('desc')}")
```

--------------------------------

### Manage Conversation Context with ChatContextManager

Source: https://context7.com/shaxiu/xianyuautoagent/llms.txt

Use ChatContextManager to persist conversation history, item details, and bargain counts in a SQLite database.

```python
from context_manager import ChatContextManager

# 初始化上下文管理器
context_mgr = ChatContextManager(
    max_history=100,  # 每个对话最多保留100条消息
    db_path="data/chat_history.db"
)

# 保存商品信息
item_info = {
    "title": "索尼降噪耳机",
    "soldPrice": 180000,  # 单位：分
    "desc": "99新头戴式降噪耳机",
    "quantity": 1,
    "skuList": [
        {"propertyList": [{"valueText": "黑色"}], "price": 180000, "quantity": 1}
    ]
}
context_mgr.save_item_info("item_123456", item_info)

# 获取商品信息
cached_item = context_mgr.get_item_info("item_123456")
print(cached_item["title"])  # 索尼降噪耳机

# 添加对话消息（基于会话ID）
chat_id = "conv_789"
context_mgr.add_message_by_chat(
    chat_id=chat_id,
    user_id="buyer_001",
    item_id="item_123456",
    role="user",
    content="这个耳机还在吗"
)

context_mgr.add_message_by_chat(
    chat_id=chat_id,
    user_id="seller_001",
    item_id="item_123456",
    role="assistant",
    content="在的亲，随时可拍"
)

# 获取对话历史
history = context_mgr.get_context_by_chat(chat_id)
print(history)
# [
#     {"role": "user", "content": "这个耳机还在吗"},
#     {"role": "assistant", "content": "在的亲，随时可拍"}
# ]

# 增加并获取议价次数
context_mgr.increment_bargain_count_by_chat(chat_id)
context_mgr.increment_bargain_count_by_chat(chat_id)
bargain_count = context_mgr.get_bargain_count_by_chat(chat_id)
print(f"议价次数: {bargain_count}")  # 议价次数: 2
```

=== COMPLETE CONTENT === This response contains all available snippets from this library. No additional content exists. Do not make further requests.