返回 MCP 目录
public公开dns本地运行

Ops监控服务器

一个服务器监控和检查工具集合,提供远程实用程序来检查服务器上的网络接口、服务状态和防火墙配置。

article

README


ops-mcp-server

中文

ops-mcp-server: an AI-driven IT operations platform that fuses LLMs and MCP architecture to enable intelligent monitoring, anomaly detection, and natural human-infrastructure interaction with enterprise-grade security and scalability.


📖 Table of Contents


🚀 Project Overview

ops-mcp-server is an IT operations management solution for the AI era. It achieves intelligent IT operations through the seamless integration of the Model Context Protocol (MCP) and Large Language Models (LLMs). By leveraging the power of LLMs and MCP's distributed architecture, it transforms traditional IT operations into an AI-driven experience, enabling automated server monitoring, intelligent anomaly detection, and context-aware troubleshooting. The system acts as a bridge between human operators and complex IT infrastructure, providing natural language interaction for tasks ranging from routine maintenance to complex problem diagnosis, while maintaining enterprise-grade security and scalability.


🌟 Key Features

🖥️ Server Monitoring

  • Real-time CPU, memory, disk inspections.
  • System load and process monitoring.
  • Service and network interface checks.
  • Log analysis and configuration backup.
  • Security vulnerability scans (SSH login, firewall status).
  • Detailed OS information retrieval.

📦 Container Management (Docker)

  • Container, image, and volume management.
  • Container resource usage monitoring.
  • Log retrieval and health checks.

🌐 Network Device Management

  • Multi-vendor support (Cisco, Huawei, H3C).
  • Switch port, VLAN, and router route checks.
  • ACL security configuration analysis.
  • Optical module and device performance monitoring.

➕ Additional Capabilities

  • Extensible plugin architecture.
  • Batch operations across multiple devices.
  • Tool listing and descriptive commands.

🎬 Demo Videos

📌 Project Demo

On Cherry Studio

Demo Animation

📌 Interactive Client Demo

On Terminal

Client Demo Animation


⚙️ Installation

Ensure you have Python 3.10+ installed. This project uses uv for dependency and environment management.

1. Install UV

curl -LsSf https://astral.sh/uv/install.sh | sh

2. Set Up Virtual Environment

uv venv .venv

# Activate the environment
source .venv/bin/activate      # Linux/macOS
.\.venv\Scripts\activate       # Windows

3. Install Dependencies

uv pip install -r requirements.txt

Dependencies are managed via pyproject.toml.


🚧 Deployment

📡 SSE Remote Deployment (UV)

cd server_monitor_sse

# Install dependencies
pip install -r requirements.txt

# Start service
cd ..
uv run server_monitor_sse --transport sse --port 8000

🐳 SSE Remote Deployment (Docker Compose)

Ensure Docker and Docker Compose are installed.

cd server_monitor_sse
docker compose up -d

# Check status
docker compose ps

# Logs monitoring
docker compose logs -f

🛠️ Local MCP Server Configuration (Stdio)

Add this configuration to your MCP settings:

{
  "ops-mcp-server": {
    "command": "uv",
    "args": [
      "--directory", "YOUR_PROJECT_PATH_HERE",
      "run", "server_monitor.py"
    ],
    "env": {},
    "disabled": true,
    "autoApprove": ["list_available_tools"]
  },
  "network_tools": {
    "command": "uv",
    "args": [
      "--directory", "YOUR_PROJECT_PATH_HERE",
      "run", "network_tools.py"
    ],
    "env": {},
    "disabled": false,
    "autoApprove": []
  },
}

Note: Replace YOUR_PROJECT_PATH_HERE with your project's actual path.


💬 Interactive Client Usage

An interactive client (client.py) allows you to interact with MCP services using natural language.

1. Install Client Dependencies

uv pip install openai rich

2. Configure Client

Edit these configurations within client.py:

# Initialize OpenAI client
self.client = AsyncOpenAI(
    base_url="https://your-api-endpoint",
    api_key="YOUR_API_KEY"
)

# Set model
self.model = "your-preferred-model"

3. Run the Client

uv run client.py [path/to/server.py]

Example:

uv run client.py ./server_monitor.py

Client Commands

  • help - Display help.
  • quit - Exit client.
  • clear - Clear conversation history.
  • model <name> - Switch models.

📄 License

This project is licensed under the MIT License.


📌 Notes

  • Ensure remote SSH access is properly configured.
  • Adjust tool parameters based on actual deployment conditions.
  • This project is under active development; feedback and contributions are welcome.

help

运行方式说明

cloud

托管运行

托管运行通常表示这个 MCP Server 由服务方环境承载,用户一般按页面提供的连接方式或授权流程接入,不需要在本地长期启动一个 MCP 进程

  1. 打开服务方连接页
  2. 完成授权或复制端点
  3. 在 MCP 客户端中连接
terminal

本地运行 / 其它方式

本地运行通常需要用户在自己的电脑或服务器上安装依赖,把 server_config 复制到 MCP 客户端,并按 env_schema 补齐环境变量、密钥或其它配置

  1. 复制 server_config
  2. 安装所需依赖
  3. 补齐环境变量后重启客户端