This repository provides a secure code execution environment using Docker containers, enabling AI applications to safely run code in isolated environments with features like flexible container management and real-time logging.
This MCP server provides a secure, containerized environment for AI applications to execute code safely. It offers flexible container management using Docker, allowing custom environments and easy file transfer between the host and containers.
Key features include: flexible Docker container management, custom environment support, file operations, command execution, real-time logging, auto-updates, and multi-platform support for Linux, macOS, and Windows.
Installation instructions are provided for Linux, macOS, and Windows, including prerequisites like Docker. Quick install scripts and manual installation steps are detailed.
The server provides tools like sandbox_initialize
to create containers, copy_project
and copy_file
for file management, write_file
to write files, sandbox_exec
to execute commands, and sandbox_stop
to remove containers. A container logs resource is also available.
Security features include isolated execution using Docker containers and resource limitations.
Configuration details are provided for Claude Desktop and other AI applications, including example JSON configurations for different operating systems.
Instructions for local development and contribution are available in DEVELOPMENT.md.
The project is licensed under the MIT License.
Automata-Labs-team/code-sandbox-mcp
January 26, 2025
March 28, 2025
Go