
Terminally MCP
An MCP server that gives AI assistants the ability to create, manage, and control terminal sessions through a safe, isolated tmux environment.
README
<div align="center">
🖥️ Terminally MCP
Supercharge AI Assistants with Terminal Control Powers ⚡
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 • Quick Start • Installation • API Reference • Examples
</div>
🎯 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:
- 🚀 Execute shell commands with full output capture
- 📂 Manage multiple terminal sessions simultaneously
- 🔍 Read terminal history and scrollback buffers
- 🛡️ Isolated tmux environment that won't interfere with your existing sessions
- ⚡ Real-time command execution with timeout protection
Perfect for AI-powered development workflows, automation, system administration, and interactive coding assistance!
✨ Features
<table> <tr> <td width="50%">
🎮 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
</td> <td width="50%">
🔒 Safe & Isolated
- Dedicated tmux server instance
- No interference with user's tmux sessions
- Timeout protection for long-running commands
- Clean session management
</td> </tr> <tr> <td width="50%">
🤖 AI-Optimized
- MCP protocol compliance
- Structured JSON responses
- Error handling and recovery
- Marker-based output capture
</td> <td width="50%">
🛠️ Developer Friendly
- TypeScript with full type safety
- Comprehensive test suite
- Clean, modular architecture
- Easy to extend and customize
</td> </tr> </table>
🚀 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.
📦 Installation
Prerequisites
- Node.js v16 or higher
- tmux installed on your system
- pnpm (recommended) or npm/yarn
Install tmux
<details> <summary>🍎 macOS</summary>
brew install tmux
</details>
<details> <summary>🐧 Linux</summary>
# Ubuntu/Debian
sudo apt-get install tmux
# Fedora
sudo dnf install tmux
# Arch
sudo pacman -S tmux
</details>
Setup for AI Assistants
<details> <summary>🤖 Claude Desktop (via MCP)</summary>
Add to your Claude Desktop configuration:
{
"mcpServers": {
"terminally-mcp": {
"command": "node",
"args": ["/path/to/terminally-mcp/build/index.js"]
}
}
}
</details>
<details> <summary>⚡ Cline/Other MCP Clients</summary>
Add to your MCP client configuration:
{
"terminally-mcp": {
"command": "node",
"args": ["/path/to/terminally-mcp/build/index.js"],
"type": "stdio"
}
}
</details>
📖 API Reference
🔧 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
}
💡 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()
});
🏗️ Architecture
terminally-mcp/
├── src/
│ ├── index.ts # Entry point
│ ├── server.ts # MCP server implementation
│ ├── services/
│ │ └── tmuxManager.ts # tmux interaction layer
│ └── tools/
│ ├── definitions.ts # Tool schemas
│ └── handlers.ts # Tool implementations
├── test/ # Test suite
├── build/ # Compiled JavaScript
└── package.json
Key Design Decisions
- 🔐 Isolated tmux Server: Each instance uses a unique socket path to prevent conflicts
- 📍 Marker-Based Output Capture: Reliable command output extraction using UUID markers
- ⏱️ Timeout Protection: Configurable timeouts prevent hanging on long-running commands
- 🎯 Type Safety: Full TypeScript implementation with strict typing
🧪 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
🤝 Contributing
We love contributions! Whether it's:
- 🐛 Bug reports
- 💡 Feature requests
- 📖 Documentation improvements
- 🔧 Code contributions
Please feel free to:
- Fork the repository
- Create your feature branch (
git checkout -b feature/AmazingFeature
) - Commit your changes (
git commit -m 'Add some AmazingFeature'
) - Push to the branch (
git push origin feature/AmazingFeature
) - Open a Pull Request
📄 License
This project is licensed under the ISC License - see the LICENSE file for details.
🙏 Acknowledgments
- Built on the Model Context Protocol specification
- Powered by tmux - the terminal multiplexer
- Inspired by the need for better AI-terminal integration
🌟 Star History
If you find this project useful, please consider giving it a ⭐ on GitHub!
<div align="center">
Built with ❤️ for the AI-assisted development community
Report Bug • Request Feature • Join Discussion
</div>
Recommended Servers
playwright-mcp
A Model Context Protocol server that enables LLMs to interact with web pages through structured accessibility snapshots without requiring vision models or screenshots.
Magic Component Platform (MCP)
An AI-powered tool that generates modern UI components from natural language descriptions, integrating with popular IDEs to streamline UI development workflow.
Audiense Insights MCP Server
Enables interaction with Audiense Insights accounts via the Model Context Protocol, facilitating the extraction and analysis of marketing insights and audience data including demographics, behavior, and influencer engagement.

VeyraX MCP
Single MCP tool to connect all your favorite tools: Gmail, Calendar and 40 more.
graphlit-mcp-server
The Model Context Protocol (MCP) Server enables integration between MCP clients and the Graphlit service. Ingest anything from Slack to Gmail to podcast feeds, in addition to web crawling, into a Graphlit project - and then retrieve relevant contents from the MCP client.
Kagi MCP Server
An MCP server that integrates Kagi search capabilities with Claude AI, enabling Claude to perform real-time web searches when answering questions that require up-to-date information.

E2B
Using MCP to run code via e2b.
Neon Database
MCP server for interacting with Neon Management API and databases
Exa Search
A Model Context Protocol (MCP) server lets AI assistants like Claude use the Exa AI Search API for web searches. This setup allows AI models to get real-time web information in a safe and controlled way.
Qdrant Server
This repository is an example of how to create a MCP server for Qdrant, a vector search engine.