MCP autoadvisor server

MCP自动交易系统服务器

概述

MCP自动交易系统服务器 是为课程项目构建的自定义模型上下文协议（MCP）服务器。\ 它提供了一套工具来探索和分析以下数据集 23000辆二手车 具有品牌、型号、年份、里程、价格、燃料类型、变速器、状况和事故历史等属性。

服务器演示了 非平凡用例 通过公开数据过滤实用程序和简单的机器学习模型 价格估算.

______________________________________________________________________

数据集

• 文件： data/Updated_Car_Sales_Data.csv

• 排： 23,000

• 柱：

______________________________________________________________________

安装

克隆存储库并在Python虚拟环境中安装依赖项：

git clone https://github.com/EstebanZG999/MCP_AutoAdvisor_Server.git
cd MCP_AutoAdvisor_Server
python -m venv .venv
source .venv/bin/activate
pip install -r requirements.txt

要求：

pandas==2.2.2
scikit-learn==1.5.2
numpy==1.26.4
mcp>=1.12,<2

______________________________________________________________________

运行服务器

直接使用以下命令运行服务器：

python server.py

通常，服务器不应手动启动—— MCP主机 根据配置通过stdio启动它。

______________________________________________________________________

MCP配置（主机）

在你的 MCP主机 存储库，将以下条目添加到 servers.yaml:

servers:
  auto_advisor:
    command: "/absolute/path/to/MCP_AutoAdvisor_Server/.venv/bin/python"
    args: ["/absolute/path/to/MCP_AutoAdvisor_Server/server.py"]
    env: {}

______________________________________________________________________

工具

此服务器通过MCP公开5个工具：

1. filter_cars

说明： 根据品牌、型号、年份、价格、里程、燃料、变速器、状况、事故等标准过滤汽车。

输入架构：

{
  "Car Make": "Toyota",
  "Year_min": 2019,
  "Price_max": 25000,
  "Transmission": "Automatic",
  "Condition": "Used",
  "Accident": "No",
  "limit": 5
}

示例调用：

/mcp call auto_advisor filter_cars {...}

______________________________________________________________________

2. recommend

说明： 在预算和偏好范围内推荐汽车。按价格升序排列。

示例输入：

{
  "budget_max": 20000,
  "Fuel Type": "Gasoline",
  "Transmission": "Automatic",
  "Condition": "Used",
  "Accident": "No",
  "Year_min": 2017,
  "limit": 5
}

______________________________________________________________________

3. estimate_price

说明： 使用线性回归模型估算汽车的价格。

示例输入：

{
  "Car Make": "Honda",
  "Car Model": "Civic",
  "Year": 2020,
  "Mileage": 40000,
  "Fuel Type": "Gasoline",
  "Transmission": "Automatic",
  "Condition": "Like New",
  "Accident": "No"
}

输出示例：

{
  "input": {...},
  "estimated_price": 28188.75
}

______________________________________________________________________

4. average_price

说明： 计算按品牌、型号、燃料类型、年份范围等过滤的汽车的平均价格。

示例输入：

{
  "Car Make": "BMW",
  "Fuel Type": "Diesel",
  "Year_min": 2018
}

______________________________________________________________________

5. top_cars

说明： 返回按价格（便宜或昂贵）排序的前N辆车。

示例输入：

{
  "n": 5,
  "sort_order": "expensive",
  "Car Make": "Audi"
}

______________________________________________________________________

演示

从主机CLI：

/mcp tools auto_advisor
/mcp call auto_advisor filter_cars {"Car Make":"Toyota","Year_min":2019,"Price_max":25000,"Transmission":"Automatic","Condition":"Used","Accident":"No","limit":5}
/mcp call auto_advisor estimate_price {"Car Make":"Honda","Car Model":"Civic","Year":2020,"Mileage":40000,"Fuel Type":"Gasoline","Transmission":"Automatic","Condition":"Like New","Accident":"No"}
/mcp call auto_advisor average_price {"Car Make":"BMW","Fuel Type":"Diesel","Year_min":2018}
/mcp call auto_advisor top_cars {"n":5,"sort_order":"expensive","Car Make":"Audi"}
/mcp call auto_advisor recommend {"budget_max":20000,"Fuel Type":"Gasoline","Transmission":"Automatic","Condition":"Used","Accident":"No","Year_min":2017,"limit":5}

______________________________________________________________________

项目结构

MCP_AutoAdvisor_Server/
├── README.md
├── requirements.txt
├── data_check.py
├── server.py              # MCP entrypoint: defines tools, routes calls, stdio runner
├── mcp_server/
│   ├── __init__.py
│   └── tools.py           # Implementation of filtering, recommendation, ML price estimator
└── data/
    └── Updated_Car_Sales_Data.csv

______________________________________________________________________