Sandbox
Executes code and configurations from Large Language Models (LLMs) in secure, isolated Docker containers.
Sandbox provides a secure environment for LLMs to execute code by leveraging Docker containers and the Model Context Protocol (MCP). It addresses the risks associated with running potentially untested code generated by LLMs directly on a user's machine. By offering an easy-to-use execution environment configured through a local, AI-native MCP server, Sandbox enables LLMs to experiment and run code safely.
Características Principales
019 GitHub stars
02Runs code in isolated Docker containers.
03Allows creation of custom sandboxes with Dockerfiles and configuration files.
04Provides strict security and network constraints for sandboxed environments.
05Supports multiple sandbox environments (shell, Python, JavaScript, Go, network-tools).
06Integrates with LLMs via the Model Context Protocol (MCP).
Casos de Uso
01Safely executing code generated by LLMs.
02Providing a controlled environment for LLMs to interact with system resources.
03Experimenting with different configurations without affecting the host system.