Security

Security MCP Server provides a production-ready Model Context Protocol implementation that lets AI agents securely access and orchestrate network security tools. It enforces strict input validation, resource limits, and observability so you can safely run automated scans, audits, and monitoring tasks in controlled environments.

How to use

You use the server by connecting an MCP client to either the HTTP API or the local stdio transport. In HTTP mode you interact via JSON over REST, monitor health and metrics, and execute tools remotely. In stdio mode you connect Claude-like clients through standard input and output streams, enabling seamless command and result exchange.

To perform a typical workflow, start the server in HTTP mode, verify the health endpoint, list available tools, then send tool execution requests. The server will validate targets, enforce security controls, route the request through a circuit breaker when needed, execute the tool, and return a structured result along with execution time and any output.

How to install

Prerequisites you need before installation are Python 3.11 or higher, pip, and nmap for the Nmap tool. You should also have a suitable environment for running Python virtual environments.

# 1. Create a project directory and set up a virtual environment
python3.11 -m venv venv
source venv/bin/activate  # On Windows: venv\Scripts\activate

# 2. Install dependencies and the MCP package
pip install --upgrade pip setuptools wheel
pip install -r requirements.txt
pip install -e .

# 3. Ensure system tools are installed
# Ubuntu/Debian
# sudo apt-get update
# sudo apt-get install -y nmap

# macOS (Homebrew)
# brew install nmap

# 4. Run in HTTP mode for API usage
export MCP_SERVER_TRANSPORT=http
export MCP_SERVER_PORT=8080
python -m mcp_server.server

# 5. Optional: run in stdio mode for Claude Desktop integration
# export MCP_SERVER_TRANSPORT=stdio
# python -m mcp_server.server

# 6. Validate health and tools endpoints
# Health: curl http://localhost:8080/health
# Tools: curl http://localhost:8080/tools
"}{

Additional content

Configuration and security are designed to be explicit and auditable. You can control how targets are validated, how inputs are sanitized, and how resources are limited. You can also enable a circuit breaker to gracefully degrade service during failures, and you can observe health and metrics to ensure ongoing reliability.

Environment variables you may configure include MCP_SERVER_TRANSPORT, MCP_SERVER_PORT, and security-related flags. The server supports multi-source configuration with hot-reload capability so you can adjust values without restarting the service.

Examples of typical usage patterns and integration steps include adding Claude Desktop integration to run the server in stdio mode, or using the HTTP API from your automation or monitoring stack. When integrating with Claude Desktop, you point the client at the stdio transport and provide the appropriate command to start the MCP server subprocess along with necessary environment variables.

If you need to test safety controls, start with non-intrusive scans and private network targets, then gradually enable more permissions in a controlled environment. Always review your security settings before moving to production.

Available tools

NmapTool

Network scanning tool with host discovery, port scanning, service/version detection, and safe script execution.

PingTool

Simple tool for safe network connectivity checks using ICMP, with input sanitization and rate limits.

TracerouteTool

Advanced tool that traces the path to a target, enforces max hops, and parses hop information into structured results.