NightTrek

๐Ÿ–ฅ๏ธ Terminally MCP

Community NightTrek
Updated

An MCP server which gives Agents the ability to use Bash Terminals using TMUX

๐Ÿ–ฅ๏ธ Terminally MCP

Supercharge AI Assistants with Terminal Control Powers โšก

MCP ProtocolNode.jsTypeScriptLicense: ISC

Give your AI assistant the power to create, manage, and control terminal sessions like a pro developer! Built on the Model Context Protocol (MCP) for seamless integration with AI tools like Claude, ChatGPT, and more.

Features โ€ข Quick Start โ€ข Installation โ€ข API Reference โ€ข Examples

๐ŸŽฏ What is Terminally MCP?

Terminally MCP is a Model Context Protocol (MCP) server that bridges the gap between AI assistants and terminal operations. It provides a safe, controlled environment where AI can:

  • ๐Ÿš€ Execute shell commands with full output capture
  • ๐Ÿ“‚ Manage multiple terminal sessions simultaneously
  • ๐Ÿ” Read terminal history and scrollback buffers
  • ๐Ÿ›ก๏ธ Isolated tmux environment that won't interfere with your existing sessions
  • โšก Real-time command execution with timeout protection

Perfect for AI-powered development workflows, automation, system administration, and interactive coding assistance!

โœจ Features

๐ŸŽฎ Terminal Control

  • Create and manage multiple terminal tabs
  • Execute commands with proper quote/escape handling
  • Capture command output and exit codes
  • Read terminal history and scrollback

๐Ÿ”’ Safe & Isolated

  • Dedicated tmux server instance
  • No interference with user's tmux sessions
  • Timeout protection for long-running commands
  • Clean session management

๐Ÿค– AI-Optimized

  • MCP protocol compliance
  • Structured JSON responses
  • Error handling and recovery
  • Marker-based output capture

๐Ÿ› ๏ธ Developer Friendly

  • TypeScript with full type safety
  • Comprehensive test suite
  • Clean, modular architecture
  • Easy to extend and customize

๐Ÿš€ Quick Start

Get up and running in under 2 minutes!

# Clone the repository
git clone https://github.com/yourusername/terminally-mcp.git
cd terminally-mcp

# Install dependencies (we recommend pnpm for speed!)
pnpm install

# Build the TypeScript code
pnpm build

# Start the MCP server
pnpm start

That's it! The server is now ready to accept MCP connections.

๐Ÿ“ฆ Installation

Prerequisites

  • Node.js v16 or higher
  • tmux installed on your system
  • pnpm (recommended) or npm/yarn

Install tmux

๐ŸŽ macOS
brew install tmux
๐Ÿง Linux
# Ubuntu/Debian
sudo apt-get install tmux

# Fedora
sudo dnf install tmux

# Arch
sudo pacman -S tmux

Setup for AI Assistants

๐Ÿค– Claude Desktop (via MCP)

Add to your Claude Desktop configuration:

{
  "mcpServers": {
    "terminally-mcp": {
      "command": "node",
      "args": ["/path/to/terminally-mcp/build/index.js"]
    }
  }
}
โšก Cline/Other MCP Clients

Add to your MCP client configuration:

{
  "terminally-mcp": {
    "command": "node",
    "args": ["/path/to/terminally-mcp/build/index.js"],
    "type": "stdio"
  }
}

๐Ÿ“– API Reference

๐Ÿ”ง Available Tools

create_tab

Creates a new terminal session.

// Request
{
  "name": "my-session"  // Optional: custom name for the tab
}

// Response
{
  "window_id": "@1"  // Unique identifier for the created tab
}
execute_command

Run any shell command in a specific terminal tab.

// Request
{
  "window_id": "@1",
  "command": "echo 'Hello, World!' && ls -la",
  "timeout": 5000  // Optional: timeout in ms (default: 10000)
}

// Response
{
  "output": "Hello, World!\ntotal 64\ndrwxr-xr-x  10 user  staff  320 Jan 15 10:00 ."
}
list_tabs

Get all active terminal sessions.

// Response
{
  "tabs": [
    {
      "window_id": "@0",
      "name": "default",
      "active": true
    },
    {
      "window_id": "@1",
      "name": "my-session",
      "active": false
    }
  ]
}
read_output

Read the terminal buffer including history.

// Request
{
  "window_id": "@1",
  "history_limit": 100  // Optional: number of history lines
}

// Response
{
  "content": "$ echo 'Previous command'\nPrevious command\n$ ls\nfile1.txt file2.txt"
}
close_tab

Close a terminal session.

// Request
{
  "window_id": "@1"
}

// Response
{
  "success": true
}

๐Ÿ’ก Examples

Basic Command Execution

// Create a new terminal
const tab = await mcp.call('create_tab', { name: 'dev-server' });

// Navigate and start a development server
await mcp.call('execute_command', {
  window_id: tab.window_id,
  command: 'cd /my/project && npm run dev'
});

// Check the output
const output = await mcp.call('read_output', {
  window_id: tab.window_id
});

Multi-Tab Workflow

// Create tabs for different purposes
const webTab = await mcp.call('create_tab', { name: 'web-server' });
const dbTab = await mcp.call('create_tab', { name: 'database' });
const testTab = await mcp.call('create_tab', { name: 'tests' });

// Start services in parallel
await Promise.all([
  mcp.call('execute_command', {
    window_id: webTab.window_id,
    command: 'npm run dev'
  }),
  mcp.call('execute_command', {
    window_id: dbTab.window_id,
    command: 'docker-compose up postgres'
  })
]);

// Run tests
await mcp.call('execute_command', {
  window_id: testTab.window_id,
  command: 'npm test'
});

Complex Command Chains

// Execute multiple commands with proper escaping
await mcp.call('execute_command', {
  window_id: '@1',
  command: `
    echo "Setting up environment..." &&
    export NODE_ENV=development &&
    echo "Installing dependencies..." &&
    npm install &&
    echo "Running migrations..." &&
    npm run migrate &&
    echo "Starting application..." &&
    npm start
  `.trim()
});

๐Ÿ—๏ธ Architecture

terminally-mcp/
โ”œโ”€โ”€ src/
โ”‚   โ”œโ”€โ”€ index.ts              # Entry point
โ”‚   โ”œโ”€โ”€ server.ts             # MCP server implementation
โ”‚   โ”œโ”€โ”€ services/
โ”‚   โ”‚   โ””โ”€โ”€ tmuxManager.ts    # tmux interaction layer
โ”‚   โ””โ”€โ”€ tools/
โ”‚       โ”œโ”€โ”€ definitions.ts    # Tool schemas
โ”‚       โ””โ”€โ”€ handlers.ts       # Tool implementations
โ”œโ”€โ”€ test/                     # Test suite
โ”œโ”€โ”€ build/                    # Compiled JavaScript
โ””โ”€โ”€ package.json

Key Design Decisions

  • ๐Ÿ” Isolated tmux Server: Each instance uses a unique socket path to prevent conflicts
  • ๐Ÿ“ Marker-Based Output Capture: Reliable command output extraction using UUID markers
  • โฑ๏ธ Timeout Protection: Configurable timeouts prevent hanging on long-running commands
  • ๐ŸŽฏ Type Safety: Full TypeScript implementation with strict typing

๐Ÿงช Development

# Run in development mode (auto-rebuild)
pnpm dev

# Run tests
pnpm test

# Run tests with UI
pnpm test:ui

# Build for production
pnpm build

# Start production server
pnpm start

๐Ÿค Contributing

We love contributions! Whether it's:

  • ๐Ÿ› Bug reports
  • ๐Ÿ’ก Feature requests
  • ๐Ÿ“– Documentation improvements
  • ๐Ÿ”ง Code contributions

Please feel free to:

  1. Fork the repository
  2. Create your feature branch (git checkout -b feature/AmazingFeature)
  3. Commit your changes (git commit -m 'Add some AmazingFeature')
  4. Push to the branch (git push origin feature/AmazingFeature)
  5. Open a Pull Request

๐Ÿ“„ License

This project is licensed under the ISC License - see the LICENSE file for details.

๐Ÿ™ Acknowledgments

  • Built on the Model Context Protocol specification
  • Powered by tmux - the terminal multiplexer
  • Inspired by the need for better AI-terminal integration

๐ŸŒŸ Star History

If you find this project useful, please consider giving it a โญ on GitHub!

Built with โค๏ธ for the AI-assisted development community

Report Bug โ€ข Request Feature โ€ข Join Discussion

MCP Server ยท Populars

MCP Server ยท New

    docmancer

    Docmancer

    Find out what your coding agents already know. Docmancer indexes the memory, rules, and instructions Claude Code, Codex, Cursor, and Gemini wrote on your machine, then carries the durable parts to every agent. Local-first, MIT.

    Community docmancer
    lineai-intelligence

    codelogic-mcp-server

    An MCP Server to utilize Codelogic's rich software dependency data in your AI programming assistant.

    HUANGCHIHHUNGLeo

    claude-real-video

    Let Claude (or any LLM) actually watch a video โ€” scene-aware, deduplicated frames + transcript, from a URL or local file. Runs locally, MIT.

    Community HUANGCHIHHUNGLeo
    jgraph

    Draw.io MCP Server

    MCP Server

    Community jgraph
    ninetails-io

    gnucash-mcp

    MCP server enabling AI assistants to read and write GnuCash accounting data

    Community ninetails-io