Claude MCP Server Collection

You can extend Claude with multiple MCP servers that run locally to perform data processing, screen interaction, data analysis, and HTTP-like operations. Each server runs as its own process and exposes capabilities you can access from Claude Desktop or other MCP clients. This guide shows practical steps to install, run, and use the collection of servers described here.

How to use

To use these MCP servers, start the ones you need on your machine and connect your Claude MCP client to them using the local endpoints defined below. Each server operates in its own process, and you can run several in parallel. In Claude Desktop, configure the corresponding MCP servers to point at the local runtimes, then invoke the available endpoints or actions from your workflows.

How to install

Prerequisites: Python 3.8 or higher, Node.js and npm (for filesystem-related functionality), Claude Desktop Application, Git.

git clone https://github.com/syedazharmbnr1/ClaudeMCPServer.git
cd ClaudeMCPServer

python3 -m venv .env
source .env/bin/activate  # On Windows: .env\Scripts\activate

pip install -r requirements.txt

# Start individual servers as needed (examples shown below) and monitor their logs for activity and errors.

Start the servers you want to use. Each command runs in its own terminal window or background process.

# DuckDB Server (data analysis)
cd fastapi/duckdb
python main.py

# Screen Capture Server (screenshots)
cd -  # return to project root
python mcp_screen_server.py

# Computer Control Server (mouse/keyboard automation)
python ComputerUse/mcp_computer_server.py

# FastAPI Integration Server (data processing API)
cd fastapi
export PYTHONPATH=/path/to/mcp-server-py
export PORT=8000
python main.py

# Curl Server (HTTP-like client functionality)
cd Curl_Server
./start_server.sh

Additional configuration and usage notes

Configure each MCP server in your Claude Desktop setup to point to the correct runtime. The following examples illustrate how to reference the local DuckDB and Stream/Screen servers from your Claude configuration.

{
  "mcpServers": {
    "duckdb": {
      "command": "python",
      "args": ["main.py"],
      "cwd": "/path/to/ClaudeMCPServer/fastapi/duckdb",
      "env": {
        "PYTHONPATH": "/path/to/ClaudeMCPServer",
        "PORT": "8010"
      }
    },
    "screen_capture": {
      "command": "python",
      "args": ["mcp_screen_server.py"],
      "cwd": "/path/to/ClaudeMCPServer",
      "env": {}
    },
    "computer_control": {
      "command": "python",
      "args": ["ComputerUse/mcp_computer_server.py"],
      "cwd": "/path/to/ClaudeMCPServer",
      "env": {}
    },
    "fastapi_integration": {
      "command": "python",
      "args": ["main.py"],
      "cwd": "/path/to/ClaudeMCPServer/fastapi",
      "env": {
        "PYTHONPATH": "/path/to/ClaudeMCPServer",
        "PORT": "8000"
      }
    },
    "curl": {
      "command": "./start_server.sh",
      "args": [],
      "cwd": "/path/to/ClaudeMCPServer/Curl_Server",
      "env": {}
    }
  }
}

Security and troubleshooting

Configure each server with the minimum necessary filesystem access. Use separate runtime environments for isolation when possible. If you encounter issues, check each server’s log files for detailed messages.

Available tools

DuckDB分析

Fast SQL querying and analysis on large CSV datasets using DuckDB in the DuckDB Server.

Screen capture

Real-time screen capture with dynamic compression and WebP support.

Computer control

Mouse/keyboard automation including text input and clipboard operations.

FastAPI integration

Robust API interface for data processing and integration tasks.

Curl server

HTTP-style client capabilities to perform local HTTP-like requests.