Enterprise SQL & Postgres C# MCP Server
Hardened C# Model Context Protocol (MCP) server for MS SQL Server and PostgreSQL databases featuring read-only transaction safeguards, command security filters, and path traversal protection.
README
🔌 Enterprise AI Toolkit (C# MCP Server)
A high-performance C# Model Context Protocol (MCP) server for Microsoft SQL Server and PostgreSQL. Connect Claude Desktop, Cursor IDE, and custom AI workflows to your production databases and filesystems safely, securely, and in under 5 minutes.
🔴 The Database Access Problem for AI
AI agents like Claude and Cursor are incredibly smart, but they can't see your data. Giving them direct access to your database is extremely risky:
- Accidental Deletions/Writes: The AI might generate
DROP TABLE,DELETE, orUPDATEcommands that corrupt production data. - Slow Schema Discovery: Manually copy-pasting tables and structures into prompts is tedious and quickly exceeds context window limits.
- Runaway Queries: An AI writing a raw
SELECT *on a table with 10 million rows will freeze your database server. - Path Traversal Attacks: Giving the AI filesystem tools can allow it to escape sandbox limits and read sensitive system files.
🟢 The Solution: Hardened C# MCP Server
This C# .NET 8.0 server provides a secure, read-only, and sandboxed bridge:
- Autonomous Schema Crawlers: The AI discovers tables, row counts, columns, primary keys, and types dynamically on-the-fly.
- Automatic Transaction Rollbacks: Every SQL query and stored procedure runs inside an explicit transaction that is automatically rolled back (
transaction.Rollback()). Writes are physically impossible. - Command Keyword Filter: Pre-scans inputs and rejects any command containing modification keywords (
INSERT,UPDATE,DELETE,DROP,ALTER,TRUNCATE,EXEC). - Row Guards: Automatically caps all query outputs with
TOP 100(SQL Server) orLIMIT 100(PostgreSQL) if no limits are specified. - Sandboxed Filesystem: Traversal-proof file tools (read, write, copy, move, delete) restricted to a configured sandbox workspace.
📦 Getting Started (Trial Guide)
This repository contains the Pre-Compiled Trial Binary of the server.
Step 1: Download & Extract
Clone this repository or download it as a ZIP and extract it to a folder (e.g., C:\Tools\EnterpriseMcpServer).
Step 2: Configure appsettings.json
Open appsettings.json in the extracted folder and configure your SQL database connection strings and sandbox directory:
{
"ConnectionStrings": {
"DefaultConnection": "Server=localhost;Database=HeroERP;Trusted_Connection=True;TrustServerCertificate=True;",
"PostgresConnection": "Host=localhost;Port=5432;Database=HeroERP;Username=postgres;Password=YOUR_PASSWORD;"
},
"McpSettings": {
"AllowedWorkspace": "./Workspace"
}
}
Step 3: Integrate with Claude Desktop
Open %APPDATA%\Claude\claude_desktop_config.json and register the server:
{
"mcpServers": {
"enterprise-sql-filesystem": {
"command": "dotnet",
"args": [
"C:\\Tools\\EnterpriseMcpServer\\bin\\EnterpriseMcpServer.dll"
]
}
}
}
Note: Make sure to replace C:\\Tools\\EnterpriseMcpServer with your actual absolute path.
Restart Claude. You will see the tools icon (hammer) active, exposing all SQL and sandboxed file tools!
🔒 Trial Version vs Full Version
The pre-compiled DLL in this repository is a Trial version. It has a limit of 10 database/file tool calls per session. If exceeded, it asks you to purchase a license.
To download the Full, Unprotected C# Source Code, Visual Studio Solution, and Unlimited Production Build, purchase the toolkit:
👉 Purchase Enterprise AI Toolkit on Gumroad
Includes lifetime updates, commercial use license, and full source code (no obfuscation, no limits).
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。