Commander MCP

🚀 commander-mcp

A Model Context Protocol (MCP) server for background process management. It enables running, monitoring, and managing background processes via PMDaemon using the MCP protocol.

✨ Features

• MCP Protocol Support: Built on the official Rust MCP SDK (v0.7)
• PMDaemon Integration: Powerful process management with built-in monitoring and log management
• Process Management Tools: Run, kill, restart, and monitor processes
• Persistent Logs: Process output is saved to files and can be read at any time
• Cross-Platform: Works on Linux, Windows, and macOS
• Standard I/O Transport: Communicates via stdin/stdout for easy integration
• Async Runtime: Built on Tokio for high-performance async operations
• Type-Safe: Leverages Rust's type system for reliability

🚀 Quick Start

Prerequisites

• Rust 1.70 or later (for building from source)
• Cargo (comes with Rust)

📦 Installation

Using the Install Script (Recommended)

The easiest way to install commander-mcp is using the install script:

curl -fsSL https://raw.githubusercontent.com/towry/commander-mcp/main/install.sh | bash

Or download and inspect the script first:

curl -fsSL https://raw.githubusercontent.com/towry/commander-mcp/main/install.sh -o install.sh
chmod +x install.sh
./install.sh

Supported platforms:

• macOS Apple Silicon (M1/M2/M3) - aarch64-apple-darwin
• Linux x86_64 - x86_64-unknown-linux-gnu

The script will:

1. Detect your platform automatically
2. Download the latest release binary
3. Install it to /usr/local/bin
4. Verify the installation

Building from Source

# Clone the repository
git clone https://github.com/towry/commander-mcp.git
cd commander-mcp

# Build the project
cargo build --release

# Run the server
cargo run --release

💻 Usage

The MCP server communicates via standard input/output using the MCP protocol. You can test it using the MCP Inspector:

# Install MCP Inspector (requires Node.js)
npm install -g @modelcontextprotocol/inspector

# If installed via install script
npx @modelcontextprotocol/inspector commander-mcp

# If running from source
npx @modelcontextprotocol/inspector cargo run --release

Using with Claude Desktop

To use this MCP server with Claude Desktop, add the following to your Claude Desktop configuration file:

macOS: ~/Library/Application Support/Claude/claude_desktop_config.json Linux: ~/.config/Claude/claude_desktop_config.json

{
  "mcpServers": {
    "commander": {
      "command": "/usr/local/bin/commander-mcp"
    }
  }
}

Or if running from source:

{
  "mcpServers": {
    "commander": {
      "command": "cargo",
      "args": ["run", "--release", "--manifest-path", "/path/to/commander-mcp/Cargo.toml"]
    }
  }
}

💻 Usage Examples

Available Tools

run

Basic Usage

Execute a command in the background.

Description: Run a command in the background. The tool waits 1 second after starting to detect early failures. Returns the process ID that can be used with other tools, along with initial logs.

Parameters:

• command (string): The command to run

Example:

{
  "command": "npm run dev"
}

Response (JSON):

{
  "process_id": "npm_run",
  "command": "npm run dev",
  "message": "Started process 'npm_run'",
  "logs": "... initial log output (if available) ..."
}

Error Response (if process fails immediately):

{
  "error": "Process 'npm_run' failed to start. Logs:\n... error logs ..."
}

Notes:

• The tool waits 1 second after starting the process to detect early failures
• If the process enters an errored or stopped state within this time, an error is returned
• Initial logs (up to 100 lines) are included in the response when available
• This helps detect issues like "address already in use" without needing to call the read tool separately

kill

Kill a previously started process by its ID.

Parameters:

• process_id (string): The process ID to kill

Example:

{
  "process_id": "npm_run"
}

Response (JSON):

{
  "process_id": "npm_run",
  "message": "Stopped process 'npm_run'"
}

kill_all

Kill all running processes.

Parameters: None

Example: Call with no parameters to stop all processes.

Response (JSON):

{
  "stopped_count": 3,
  "message": "Stopped 3 process(es)"
}

read

Read the stdout output from a process by its ID.

Description: Returns recent log output (default: last 1000 lines).

Parameters:

• process_id (string): The process ID to read output from
• lines (number, optional): Maximum number of log lines to return (default: 1000)

Example:

{
  "process_id": "npm_run",
  "lines": 500
}

Response (JSON):

{
  "process_id": "npm_run",
  "logs": "... log output here ..."
}

restart

Restart a previously run process by its ID.

Parameters:

• process_id (string): The process ID to restart

Example:

{
  "process_id": "npm_run"
}

Response (JSON):

{
  "process_id": "npm_run",
  "message": "Restarted process 'npm_run'"
}

list

List all currently running or stopped processes.

Parameters: None

Response (JSON):

{
  "processes": [
    {
      "name": "npm_run",
      "status": "running",
      "pid": 12345,
      "started_at": "2025-10-03 17:14:55 UTC",
      "uptime_seconds": 120,
      "restarts": 0,
      "cpu_usage": 2.5,
      "memory_mb": 45
    }
  ]
}

Returns a list of all processes with their status, PID, started_at timestamp, uptime, CPU usage, memory usage, and restart count.

📚 Documentation

How It Works

1. Process Management: Uses PMDaemon, a high-performance process manager written in Rust
2. Process IDs: Automatically generated based on command keywords (e.g., npm run dev → npm_run)
3. Persistent Logs: All process output is saved to log files that can be read at any time
4. Real-time Monitoring: Track CPU, memory usage, and process state
5. Cross-platform: Native support for Linux, Windows, and macOS

Key Features

• Persistent log management: Logs are saved to files and can be read at any time (not just live output)
• Per-process log access: Can read logs for specific processes
• No external dependencies: PMDaemon is a Rust library, no external binaries needed
• Cross-platform: Native support for Windows, macOS, and Linux
• Built-in monitoring: CPU and memory tracking included
• Direct API: Programmatic control through Rust API

🔧 Technical Details

Development

Running Tests

# Run all tests
cargo test --all-features --verbose

# Run tests with output
cargo test -- --nocapture

Code Quality

This project follows Rust best practices and uses conventional commits for automated releases.

# Format code (REQUIRED before committing)
cargo fmt --all

# Check formatting
cargo fmt --all -- --check

# Run linting
RUSTFLAGS="-D warnings" cargo clippy --workspace --all-targets --all-features --verbose

Using Just Commands

This project includes a justfile for convenient development:

# Format, lint, and test
just check

# Format code
just format

# Run linting
just lint

# Run tests
just test

# Build release binary
just build

Project Structure

commander-mcp/
├── src/
│   ├── main.rs              # Entry point with MCP server setup
│   ├── process_server.rs    # MCP tools implementation
│   └── process_manager.rs   # PMDaemon wrapper for process management
├── Cargo.toml               # Project dependencies and metadata
└── README.md                # This file

Related

• PMDaemon - The process manager library powering this server
• MCP Protocol - Model Context Protocol specification
• Rust MCP SDK - Official MCP SDK for Rust