MCP Agentic Terminal

MCP Agentic Terminal is a lightweight terminal automation server built with FastMCP that exposes powerful local system tools through the Model Context Protocol (MCP).

It enables AI agents and MCP-compatible clients to safely interact with the local operating system by executing shell commands, running Python code, managing files, and performing filesystem operations programmatically.

Features

• Execute bash commands
• Run inline Python code
• Execute Python files
• Read and write files
• Create and delete directories
• List directory contents
• Search files using glob patterns
• Search text patterns inside files
• Delete files and folders
• STDIO-based MCP transport support

Included Tools

Terminal Execution

• bash(command)

  • Execute shell commands and return stdout/stderr.

Python Execution

• python_code(code)

  • Execute inline Python code.

• python_file(file)

  • Execute Python scripts directly.

File Operations

• read_file(file)

  • Read file contents.

• write_file(file, content)

  • Create or overwrite files.

• delete_file(file)

  • Remove files.

Directory Operations

• create_directory(directory)

  • Create directories recursively.

• list_directory(directory)

  • List files and folders.

• delete_folder(folder)

  • Delete folders recursively.

Search Utilities

• glob(pattern)

  • Match filesystem patterns.

• grep(pattern, file)

  • Search for text patterns inside files.

Usage

Run the MCP server:

python main.py

Or:

uv run main.py

The server uses STDIO transport by default and can be connected to any MCP-compatible client.

Example MCP Client Configuration

from langchain_mcp_adapters.client import MultiServerMCPClient

client = MultiServerMCPClient(
    {
        "mcp_agentic_terminal": {
            "transport": "stdio",
            "command": "uvx",
            "args": ["mcp_agentic_terminal"],
        }
    }
)

Use Cases

• AI coding assistants
• Autonomous development agents
• Local task automation
• File system orchestration
• AI-powered DevOps workflows
• Secure local execution environments

Tech Stack

• FastMCP
• Python
• MCP (Model Context Protocol)

License

MIT License