Skip to main content
Glama
B-DINESH9

MCP System Tools Server

by B-DINESH9
README.md
# MCP System Tools Server

This is a production-ready **Model Context Protocol (MCP)** server built in Python. 
It provides custom tools that AI Agents (like Claude or custom LangChain/LlamaIndex bots) can connect to and use to interact with the host system.

## ๐Ÿ› ๏ธ Included Tools
1. `get_system_stats`: Retrieves current live system statistics (CPU usage, Memory usage, Available Memory).
2. `calculate_expression`: A safe math evaluator that performs basic arithmetic operations.

## ๐Ÿš€ How to Run (Docker - Recommended)
If you have Docker installed, you can spin up the entire isolated environment with one command:
```bash
docker-compose up --build -d
```
The Server-Sent Events (SSE) endpoint will be available at: `http://localhost:8000/sse`

## ๐Ÿ’ป How to Run (Local Python)
If you don't have Docker, you can run the server directly using Python.

1. Install the requirements:
```bash
pip install -r requirements.txt
```
2. Run the server:
```bash
python mcp_server.py
```
The server will boot up and listen for connections on `http://127.0.0.1:8000/sse`.

## ๐Ÿงช Testing the Server
You can quickly test your tools using the official MCP Inspector. Run the following command in a new terminal:
```bash
npx -y @modelcontextprotocol/inspector
```
* Select **SSE** as the transport.
* Enter `http://127.0.0.1:8000/sse` as the URL and click **Connect**.

Alternatively, run the included `test_client.py` script:
```bash
python test_client.py
```

## ๐Ÿค– Connecting to Claude Desktop
You can easily give Claude Desktop the ability to use these tools. Add this configuration to your Claude config file (`claude_desktop_config.json`):

```json
{
  "mcpServers": {
    "system-tools": {
      "command": "python",
      "args": [
        "path/to/your/mcp_server.py"
      ]
    }
  }
}
```