The Tabby VSCode Agent is an advanced plugin for Tabby Terminal, designed to seamlessly integrate AI-powered terminal control and automation directly within VS Code. Building upon the foundations of the original Tabby MCP Server, this improved version offers enhanced functionality, quicker speeds, and a more refined user experience tailored for VS Code users.

• VSCode Extension Support, listens to http port.
• Fixed URL Opening links
• Increased speed of reading SSH sessions
• Fixed issue with opening VSCode via non extension
• Fixed UI for Command History
• Enhanced support for VS Code integration
• Added support for VSCode Extension
• Copilot agent chat can now be opened directly from the navbar
• New and improved UI elements for a more modern look
• Updated and fixed navbar for better navigation and stability
• Improved command execution speeds and responsiveness
• Quick access to settings and chat from the main interface
• More robust error handling and feedback for users
• Optimized MCP server communication for lower latency
• Improved hotkey support and customization
• Better handling of long-running commands and output

• Navbar rendering issues in some VS Code versions
• UI glitches in command dialogs and modals
• Minor bugs in command history and output storage
• Various performance and stability improvements

Witness the Tabby VSCode Agent in action with these comprehensive video demonstrations:

• 🤖 AI Integration: Effortlessly connect AI assistants to your terminal for intelligent command execution and automation.
• 🔌 Built-in MCP Server: Features a robust Model Context Protocol (MCP) server implementation for reliable communication.
• 🚀 Deep VS Code Integration:
  • Open Chat from Navbar: Instantly open the Copilot chat window directly from the Tabby navigation bar within VS Code.
  • Optimised Settings & Logins: Enjoy a redesigned settings interface and streamlined login processes for a smoother workflow.
  • Enhanced Speed: Experience significantly quicker response times and overall performance.
• 📦 Bundled Stdio Server: Includes tabby-mcp-stdio for a stable and efficient connection with VS Code.
• 🖥️ Terminal Control: Empower AI to execute commands, read terminal output, and manage your sessions.
• 🔍 Session Management: View and manage your SSH sessions directly from the plugin.
• 🚫 Command Abort: Safely abort any running commands with ease.
• 📋 Buffer Access: Retrieve terminal buffer content with flexible line range options.
• 🔒 Pair Programming Mode: An optional safety feature that requires confirmation before AI executes commands, ensuring you maintain control.
• 📊 Command History: Keep track of and review all previously executed commands.
• 🔄 Command Output Storage: Access complete command outputs with convenient pagination.

1. Open Tabby settings and navigate to Plugins → MCP.
2. Locate and click "Install" on the Tabby VSCode Agent plugin.
3. Restart Tabby to finalise the installation.
4. Proceed to configure VS Code with the MCP Server as detailed below.

1. Ensure the plugin is installed using one of the methods above.
2. Launch Tabby and go to Settings → Copilot.
3. Within the settings page, you'll find a collapsible "Instructions" section. This provides detailed, step-by-step guidance on how to integrate the agent with VS Code.
4. Configure the MCP server port (default: 3001).
5. Toggle "Start on Boot" if you wish for the server to automatically launch with Tabby.
6. Connect your preferred AI client to the MCP server. A list of supported clients can be found at https://modelcontextprotocol.io/clients.

1. Start Tabby with the Tabby VSCode Agent plugin enabled.
2. Configure your AI client to connect to the MCP server (refer to Connecting to MCP for details).
3. Instruct your AI assistant to run commands or manage your terminal sessions.

Example AI prompt:

Connect to my Tabby VSCode Agent and list all available terminal sessions.
Then execute the command "ls -la" in the first available terminal.

To configure AI clients to use your MCP server, add the following to your mcp.json file:

{
	"servers": {
		"tabby": {
			"url": "http://localhost:3001/sse",
			"type": "http"
		}
	}
}

For a more robust integration with VS Code, you can utilise the bundled stdio server.

1. Navigate to Settings → Copilot in Tabby.
2. Under the Stdio Server section, you will find the full path to the server script.
3. Click the "Copy" button to copy the path to your clipboard.
4. In your VS Code mcp.json, configure the server as shown below, pasting the copied path:

{
    "servers": {
        "tabby-stdio": {
            "type": "stdio",
            "command": "node",
            "args": ["<PASTE THE COPIED PATH HERE>"]
        }
    }
}

The plugin includes a "Pair Programming Mode" to enhance safety when AI assistants control your terminal:

• Confirmation Dialogue: Prompts the user for confirmation before executing commands.
• Auto Focus Terminal: Automatically focuses the terminal when commands are executed.
• Command Rejection: Provides the ability to reject commands with feedback.

To enable Pair Programming Mode:

1. Go to Tabby settings → Copilot.
2. Toggle "Enable Pair Programming Mode".
3. Configure additional safety options as required.

Contributions are highly encouraged! Here's how you can contribute to the Tabby VSCode Agent:

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

Please refer to the contributing guidelines for more detailed information.

1. Clone the repository and install dependencies:

   git clone https://github.com/SteffMet/tabby-vscode-agent.git
   cd tabby-vscode-agent
   npm install

2. Make your desired changes to the codebase.

3. Build the plugin:

   npm run build

4. Test the plugin with Tabby:

   npm run deploy

This project is licensed under the MIT Licence - see the LICENCE file for details.

Made with ❤️ by SteffMet