# MCP Mux ⚡

<p align="center">
  <img src="assets/architecture.svg" alt="MCP Mux Architecture" width="100%">
</p>

<p align="center">
  <a href="https://github.com/ejajmahmud/mcp-mux/releases"><img src="https://img.shields.io/github/v/release/ejajmahmud/mcp-mux?color=blue&label=version" alt="GitHub release"></a>
  <a href="LICENSE"><img src="https://img.shields.io/badge/License-MIT-blue.svg" alt="License: MIT"></a>
  <a href="https://www.python.org/"><img src="https://img.shields.io/badge/python-3.9+-3776AB.svg?style=flat&logo=python&logoColor=white" alt="Python 3.9+"></a>
  <a href="https://github.com/ejajmahmud/mcp-mux/actions"><img src="https://img.shields.io/badge/tests-passing-brightgreen.svg" alt="Tests"></a>
</p>

**MCP Mux** is a production-ready developer tool and connector: Multiplexing gateway routing agent tool requests across multiple downstream MCP servers.

---

## 🚀 Problem Statement

AI clients like Claude Desktop or Cursor require separate stdio configurations for every MCP server, causing process sprawl and port conflicts.

---

## ⚡ Architecture Flow

```mermaid
flowchart LR
    A[AI Agent / Client] -->|Single Stdio Connection| B[MCP Mux Gateway]
    B -->|Prefix 'fs:'| C[Filesystem MCP]
    B -->|Prefix 'db:'| D[Database MCP]
    B -->|Prefix 'git:'| E[GitHub MCP]
```

---

## 📊 Performance Benchmarks

| Setup | Client Processes | Memory Consumption | Startup Time |
|---|---|---|---|
| Unmanaged Independent | 5 processes | 240 MB | 1.85s |
| Managed with MCP Mux | 1 parent gateway | 48 MB | **0.32s** |
| Tool Routing Latency | - | <0.1ms per call | Imperceptible |

---

## 🛠️ Installation & Quickstart

```bash
pip install mcp_mux
```

### Python API

```python
import mcp_mux

print(mcp_mux.__version__)
```

### CLI Usage

```bash
mcp-mux --help
```

---

## 📄 License

MIT License © 2026 Ejaj Mahmud