sfgarza/network-capture-mcp

Network Capture MCP

An MCP (Model Context Protocol) server that provides HTTP(S) and WebSocket network capture and analysis tools using mockttp.

What is MCP? The Model Context Protocol (MCP) is a standard that allows AI agents to access external tools and data sources. This server provides AI agents with the ability to capture, analyze, and query network traffic from your applications using natural language.

Why Use This?

• AI-Assisted Debugging: Let AI analyze request/response patterns, headers, and payloads to identify root causes
• Natural Language Queries: Ask your AI "What API calls failed today?" or "Which endpoints are slowest?"
• Performance Analysis: AI finds slow endpoints and provides optimization recommendations
• Automated Monitoring: AI detects traffic patterns and anomalies automatically

Quick Start

1. Install and Setup

git clone https://github.com/theinfinitereality/network-capture-mcp
cd network-capture-mcp
npm install

2. Configure Your AI Agent

Add to your Claude Desktop config (claude_desktop_config.json):

{
  "mcpServers": {
    "netcap": {
      "command": "npx",
      "args": ["tsx", "/full/path/to/network-capture-mcp/src/index.ts"]
    }
  }
}

3. Configure Your Application

Point your application to use the proxy:

export HTTP_PROXY=http://localhost:8080
export HTTPS_PROXY=http://localhost:8080
your-application

4. Start Capturing

npm start

5. Ask Your AI Agent

"Show me all API requests from the last 10 minutes"
"Analyze traffic patterns and find any issues"
"What endpoints are responding slowly?"

That's it! Your AI agent can now analyze your application's network traffic.

Next Steps

1. - Detailed installation, configuration, and first usage
2. - Configure Claude Desktop and other MCP clients
3. - Real-world examples and integration patterns
4. - Complete configuration options and CLI parameters

Documentation

• - Complete MCP tools documentation with parameters and examples
• - Technical architecture and system design
• - Contributing, testing, and development setup

Need Help?

Quick Issues: See for solutions to common problems like port conflicts, setup issues, and connection problems.

Requirements

• Node.js 18+ - Download from nodejs.org
• AI Agent with MCP Support - Such as Claude Desktop
• Applications to Monitor - The apps/services you want to capture traffic from

Key Features

• HTTP/HTTPS & WebSocket Traffic Capture with complete request/response data
• AI-Powered Analysis using natural language queries
• Performance Monitoring with response time tracking and bottleneck detection
• Multiple Export Formats (JSON, CSV, HAR) with automatic health monitoring

Important: This tool is designed for development and testing only. Do not use in production environments.

License

MIT