NOC.McpServer
Exposes a tool to convert JSON samples into equivalent C# class definitions using NiceOneCode's converter. AI assistants can call directly without pasting JSON into a web form.
README
NOC.McpServer
An MCP (Model Context Protocol) server that exposes NiceOneCode's JSON → C# class converter as a tool AI assistants can call directly — no more pasting JSON into a web form.
Built with the official ModelContextProtocol C# SDK, running over stdio transport. Published on NuGet.org under the McpServer package type.
What it does
Exposes a single tool, json_to_c_sharp, which takes a JSON sample and returns the equivalent C# class definitions, generated by NiceOneCode's converter.
Authentication is handled transparently: the server logs in to your NiceOneCode account on first use, caches the JWT, and automatically re-authenticates if it expires.
Prerequisites
- A NiceOneCode account (userid + password)
- .NET 9 SDK or later, to install/run the tool
- .NET 10 SDK if you want to use
dnx(VS Code, one-shot execution) instead of a persistent install - .NET 9 SDK specifically if building from source
Creating an Account
If you don't already have a NiceOneCode account, you can register one via API:
curl -X POST \
--header 'Content-Type: application/json' \
--header 'Accept: application/json' \
-d '{
"UserName": "your-username",
"Password": "your-password",
"Email": "your-email@example.com",
"GenderID": 1
}' \
'https://www.niceonecode.com/api/nc-register'
| Field | Description |
|---|---|
UserName |
Desired username |
Password |
Desired password — use this as NOC_PASSWORD |
Email |
A valid email address |
GenderID |
1 = Male, 2 = Female |
Important: the response body is a plain string, not a JSON object — for example:
"your-userid"
Use this returned value as NOC_USERID (not necessarily assumed to always match the UserName you submitted — always use the value the API actually returns, rather than the value you sent).
Install
dotnet tool install -g NOC.McpServer
This gives you the noc-mcp command, used in every client config below.
Configuration
The server reads two required environment variables:
| Variable | Description |
|---|---|
NOC_USERID |
Your NiceOneCode userid |
NOC_PASSWORD |
Your NiceOneCode password |
Never commit real credentials. Set them via your MCP client's config (see below) or your local shell environment.
Connecting to an MCP client
Claude Desktop
Edit claude_desktop_config.json (Windows: %APPDATA%\Claude\claude_desktop_config.json, macOS: ~/Library/Application Support/Claude/claude_desktop_config.json):
{
"mcpServers": {
"niceonecode": {
"command": "noc-mcp",
"env": {
"NOC_USERID": "your-userid",
"NOC_PASSWORD": "your-password"
}
}
}
}
Fully quit and reopen Claude Desktop after editing.
Claude Code
claude mcp add-json niceonecode '{"command":"noc-mcp","env":{"NOC_USERID":"your-userid","NOC_PASSWORD":"your-password"}}'
Verify with claude mcp list.
Codex
codex mcp add niceonecode --env NOC_USERID=your-userid --env NOC_PASSWORD=your-password -- noc-mcp
Verify inside a session with /mcp.
VS Code
Add to .vscode/mcp.json in your workspace, or your global VS Code MCP settings:
{
"inputs": [
{
"type": "promptString",
"id": "NOC_USERID",
"description": "Your NiceOneCode account userid"
},
{
"type": "promptString",
"id": "NOC_PASSWORD",
"description": "Your NiceOneCode account password",
"password": true
}
],
"servers": {
"NOC.McpServer": {
"type": "stdio",
"command": "dnx",
"args": ["NOC.McpServer", "--yes"],
"env": {
"NOC_USERID": "${input:NOC_USERID}",
"NOC_PASSWORD": "${input:NOC_PASSWORD}"
}
}
}
}
VS Code will prompt for your userid and password the first time the server starts. Requires the .NET 10 SDK for dnx, which downloads and runs the latest published version on demand — no separate install step needed. Pin a specific version with "NOC.McpServer@1.0.7" instead of "NOC.McpServer" if you want reproducible behavior rather than always-latest.
Note: a global.json in your working directory pinning an SDK below .NET 10 will cause dnx to fail with a confusing "unrecognized argument" error rather than a clear version mismatch — check for one if this happens.
Gemini CLI
Edit ~/.gemini/settings.json (global) or .gemini/settings.json (project-specific):
{
"mcpServers": {
"niceonecode": {
"command": "noc-mcp",
"env": {
"NOC_USERID": "your-userid",
"NOC_PASSWORD": "your-password"
}
}
}
}
Restart Gemini CLI, then run /mcp to confirm it's connected.
Windows note: if the server fails to start directly, wrap it through cmd:
{
"mcpServers": {
"niceonecode": {
"command": "cmd",
"args": ["/c", "noc-mcp"],
"env": { "NOC_USERID": "your-userid", "NOC_PASSWORD": "your-password" }
}
}
}
Windsurf
Edit ~/.codeium/windsurf/mcp_config.json (macOS/Linux) or %USERPROFILE%\.codeium\windsurf\mcp_config.json (Windows), or open it via the MCPs icon in the Cascade panel → Configure:
{
"mcpServers": {
"niceonecode": {
"command": "noc-mcp",
"env": {
"NOC_USERID": "your-userid",
"NOC_PASSWORD": "your-password"
}
}
}
}
Fully quit and reopen Windsurf after saving.
Devin
Devin (cloud): Settings → Connections → MCP servers → Add a custom MCP:
{
"transport": "STDIO",
"command": "noc-mcp",
"args": [],
"env_variables": {
"NOC_USERID": "your-userid",
"NOC_PASSWORD": "your-password"
}
}
Devin for Terminal: add to .devin/config.json:
{
"mcpServers": {
"niceonecode": {
"command": "noc-mcp",
"env": {
"NOC_USERID": "your-userid",
"NOC_PASSWORD": "your-password"
}
}
}
}
Devin Desktop shares Windsurf's mcp_config.json — no separate setup needed if already configured above.
Building from source
If you'd rather build locally instead of installing from NuGet:
git clone https://github.com/nice-one-code/NOC.McpServer.git
cd NOC.McpServer
dotnet build
Then point any client above at the compiled DLL instead of noc-mcp, e.g. for Claude Desktop:
{
"mcpServers": {
"niceonecode": {
"command": "dotnet",
"args": ["/absolute/path/to/NOC.McpServer/bin/Debug/net9.0/NOC.McpServer.dll"],
"env": {
"NOC_USERID": "your-userid",
"NOC_PASSWORD": "your-password"
}
}
}
}
Use forward slashes in the path even on Windows. The same substitution (dotnet + DLL path in place of noc-mcp) applies to any client config in this README.
Testing locally with the MCP Inspector
npx @modelcontextprotocol/inspector dotnet bin/Debug/net9.0/NOC.McpServer.dll
This opens a browser UI where you can invoke json_to_c_sharp manually and inspect the raw request/response. Set your credentials under the Inspector's Environment Variables panel before connecting.
Project structure
推荐服务器
Baidu Map
百度地图核心API现已全面兼容MCP协议,是国内首家兼容MCP协议的地图服务商。
Playwright MCP Server
一个模型上下文协议服务器,它使大型语言模型能够通过结构化的可访问性快照与网页进行交互,而无需视觉模型或屏幕截图。
Magic Component Platform (MCP)
一个由人工智能驱动的工具,可以从自然语言描述生成现代化的用户界面组件,并与流行的集成开发环境(IDE)集成,从而简化用户界面开发流程。
Audiense Insights MCP Server
通过模型上下文协议启用与 Audiense Insights 账户的交互,从而促进营销洞察和受众数据的提取和分析,包括人口统计信息、行为和影响者互动。
VeyraX
一个单一的 MCP 工具,连接你所有喜爱的工具:Gmail、日历以及其他 40 多个工具。
graphlit-mcp-server
模型上下文协议 (MCP) 服务器实现了 MCP 客户端与 Graphlit 服务之间的集成。 除了网络爬取之外,还可以将任何内容(从 Slack 到 Gmail 再到播客订阅源)导入到 Graphlit 项目中,然后从 MCP 客户端检索相关内容。
Kagi MCP Server
一个 MCP 服务器,集成了 Kagi 搜索功能和 Claude AI,使 Claude 能够在回答需要最新信息的问题时执行实时网络搜索。
e2b-mcp-server
使用 MCP 通过 e2b 运行代码。
Neon MCP Server
用于与 Neon 管理 API 和数据库交互的 MCP 服务器
Exa MCP Server
模型上下文协议(MCP)服务器允许像 Claude 这样的 AI 助手使用 Exa AI 搜索 API 进行网络搜索。这种设置允许 AI 模型以安全和受控的方式获取实时的网络信息。