Skip to main content
Glama
TJYOYO
by TJYOYO
README.md
# MCP 项目

[MCP Official](https://github.com/modelcontextprotocol) 

[Python SDK](https://github.com/modelcontextprotocol/python-sdk)

A minimal [FastMCP](https://github.com/jlowin/fastmcp) server demo with example tools.

## Requirements

- Python 3.13+
- [uv](https://docs.astral.sh/uv/)

## init evn

```bash
uv init .
```

## 安装 MCP SDK

[MCP SDK](https://github.com/modelcontextprotocol/python-sdk?utm_source=chatgpt.com#overview) 是一个 Python 库,提供了与 MCP 服务器进行通信的接口。它允许开发者轻松地构建与 MCP 服务器交互的应用程序。

```bash
uv add "mcp[cli]"
```

## 写MCP的具体实现

1: 使用注解@mcp.tool()修饰,然后完成一个function 函数,like

```code
@mcp.tool()
def generate_leave_email(date: str) -> str:
    """Generate a leave request email for the given date (YYYY-MM-DD)."""
    return (
        f"Hi, all, I'll take a leave on {date}. "
        "If you have any questions, please let me know. Thanks! Best regards, [Your Name]"
    )
```

2: mcp启动方法

```code
if __name__ == "__main__":
    mcp.run()
```

## 添加到VS Code for Windows

方式1: 在.vscode目录下创建mcp.json文件,并添加以下内容:

```code
{
    "servers": {
        "mcp-server": {
           "type": "stdio",
           "command": "C:\\Users\\tianjin\\.local\\bin\\uv.exe",
           "args": ["run", "python", "main.py"],
           "cwd": "D:\\AI\\mcp-server"
        }
    },
    "inputs": []
}
```

1: "mcp-server": 是服务器的名称,可以自定义。

2: "type": "stdio": 指定服务器的通信方式为标准输入输出。

3: "command": "C:\\Users\\tianjin\\.local\\bin\\uv.exe": 指定要运行的命令,这里是uv.exe的路径。

4: "args": ["run", "python", "main.py"]: 指定运行命令的参数,这里是运行Python脚本main.py。

5: "cwd": "D:\\AI\\mcp-server": 指定服务器的工作目录,这里是mcp-server的路径。

方式2:通过mcp servers插件,然后命令面板(Ctrl+Shift+P)选择 "MCP: Add Server",然后按照提示输入服务器的名称、类型、命令、参数和工作目录。

## 添加到VS Code for MAC

like, another server config.

```code
"testMcp": {
           "command": "uv",
           "args": [
           "run",
              "--project",
              "/Users/jacktian/Desktop/AI/McpTest",
           "python",
           "/Users/jacktian/Desktop/AI/McpTest/main.py"
           ]
        }
```

## start server

在mcp.json中,点击start 按钮,然后就可以在copilot中使用mcp-server了。

## validation

![image](Screenshot2.png)

## 克隆后首次运行(避免 `mcp.server.fastmcp` 无法 resolve)

拉取仓库后,如果编辑器提示 `mcp.server.fastmcp` 无法解析,通常是因为还没在当前工作区安装依赖,或 VS Code 没选中项目虚拟环境。

建议按下面顺序执行:

1. 在项目根目录创建并同步依赖(会根据 `pyproject.toml` 和 `uv.lock` 安装 `mcp`):
    - `uv sync`
2. 在 VS Code 里选择解释器为项目下 `.venv`。
3. 重载窗口或重启 Python Language Server。

本仓库已包含 `.vscode/settings.json`,会优先指向 `${workspaceFolder}/.venv/bin/python`。

TDQS

A3.9/5.0

Scored across 1 tool

Disambiguation5/5

Only one tool exists, so there is no possible confusion with other tools.

Naming Consistency5/5

The single tool uses a clear verb_noun pattern ('add_numbers'), consistent with common conventions.

Tool Count3/5

With only one tool, the server feels minimal for a generic name like 'mcp-server', but it could be appropriate if the scope is intentionally narrow.

Completeness2/5

The server only provides addition, missing other basic arithmetic operations, leaving significant gaps for mathematical tasks.

Maintenance

ActivityInactive
ResponsivenessNo issues