ms-enclave

Modularized and Stable Sandbox runtime environment

Overview

ms-enclave is a modularized and stable sandbox runtime environment that provides secure isolated execution environments for applications. It supports multiple programming languages and frameworks, ensuring code runs safely in controlled environments with Docker-based containerization.

Features

🔒 Secure Isolation: Complete isolation using Docker containers
🧩 Modular Design: Plugin-based architecture with extensible tools
⚡ High Performance: Optimized runtime performance with resource monitoring
📊 Resource Monitoring: Real-time CPU, memory, and resource usage tracking
🛡️ Security Policies: Configurable security policies and permission control
🌐 HTTP API: RESTful API for remote sandbox management
🔧 Tool System: Extensible tool system for different execution environments

Requirements

Python >= 3.9
Operating System: Linux, macOS, or Windows with Docker support

Installation

Install from PyPI

pip install ms-enclave

Install from Source

git clone https://github.com/modelscope/ms-enclave.git
cd ms-enclave
pip install -e .

Quick Start

Basic Usage

import asyncio
from ms_enclave.sandbox.boxes import SandboxFactory
from ms_enclave.sandbox.model import DockerSandboxConfig, SandboxType

async def main():
    # Create Docker sandbox configuration
    config = DockerSandboxConfig(
        image='python:3.11-slim',
        timeout=30,
        memory_limit='512m',
        tools_config={'python_executor': {}}
    )

    # Create and use sandbox with context manager
    async with SandboxFactory.create_sandbox(SandboxType.DOCKER, config) as sandbox:
        # Execute Python code
        result = await sandbox.execute_tool('python_executor', {
            'code': "print('Hello from sandbox!')\nresult = 2 + 2\nprint(f'2 + 2 = {result}')"
        })
        print(f'Result: {result.output}')

asyncio.run(main())

HTTP Server Usage

from ms_enclave.sandbox import create_server

# Start the sandbox server
server = create_server(cleanup_interval=300)
server.run(host='127.0.0.1', port=8000)

python -m ms_enclave.run_server

HTTP Manager Client

import asyncio
from ms_enclave.sandbox.manager import HttpSandboxManager
from ms_enclave.sandbox.model import DockerSandboxConfig, SandboxType

async def main():

    async with HttpSandboxManager(base_url='http://127.0.0.1:8000') as manager:
        # Create sandbox
        config = DockerSandboxConfig(image='python:3.11-slim', tools_config={'python_executor': {}})
        sandbox_id = await manager.create_sandbox(SandboxType.DOCKER, config)

        # Execute code
        result = await manager.execute_tool(
            sandbox_id, 'python_executor',
            {'code': 'print("Hello from remote sandbox!")'}
        )
        print(result.model_dump())

asyncio.run(main())

API Reference

SandboxFactory

create_sandbox(sandbox_type, config)

Create a new sandbox instance.

sandbox = SandboxFactory.create_sandbox(SandboxType.DOCKER, config)

Sandbox Methods

execute_tool(tool_name, parameters)

Execute a tool within the sandbox.

result = await sandbox.execute_tool('python_executor', {
    'code': 'print("Hello World")',
    'timeout': 30
})

get_available_tools()

Get list of available tools.

tools = sandbox.get_available_tools()

start() / stop() / cleanup()

Manage sandbox lifecycle.

await sandbox.start()
await sandbox.stop()
await sandbox.cleanup()

HttpSandboxManager

Remote sandbox management via HTTP API.

create_sandbox(sandbox_type, config)

sandbox_id = await manager.create_sandbox(SandboxType.DOCKER, config)

execute_tool(sandbox_id, tool_name, parameters)

result = await manager.execute_tool(sandbox_id, 'python_executor', params)

list_sandboxes(status_filter=None)

sandboxes = await manager.list_sandboxes()

Examples

Advanced Python Execution

async def advanced_example():
    config = DockerSandboxConfig(
        image='python:3.11-slim',
        tools_config={'python_executor': {}},
        memory_limit='1g'
    )

    async with SandboxFactory.create_sandbox(SandboxType.DOCKER, config) as sandbox:
        # Data processing example
        code = '''
import json
import statistics

data = [1, 2, 3, 4, 5, 6, 7, 8, 9, 10]
result = {
    "mean": statistics.mean(data),
    "median": statistics.median(data),
    "stdev": statistics.stdev(data)
}
print(json.dumps(result, indent=2))
'''

        result = await sandbox.execute_tool('python_executor', {'code': code})
        print(result.output)

Error Handling

async def error_handling_example():
    config = DockerSandboxConfig(
        image='python:3.11-slim',
        tools_config={'python_executor': {}},
        timeout=5
    )

    async with SandboxFactory.create_sandbox(SandboxType.DOCKER, config) as sandbox:
        # Handle syntax errors
        result = await sandbox.execute_tool('python_executor', {
            'code': 'print("Missing quote'
        })

        if result.error:
            print(f"Error: {result.error}")
        else:
            print(f"Output: {result.output}")

Development

Local Development Setup

# Clone the repository
git clone https://github.com/modelscope/ms-enclave.git
cd ms-enclave

# Install dependencies
pip install -e ".[dev]"

# Run examples
python examples/usage_examples.py
python examples/server_example.py

Contributing

We welcome contributions! Please see CONTRIBUTING.md for details.

Contributing Steps

Fork the repository
Create a feature branch: git checkout -b feature/amazing-feature
Make your changes and add tests
Run tests: pytest
Commit changes: git commit -m 'Add amazing feature'
Push to branch: git push origin feature/amazing-feature
Submit a Pull Request

License

This project is licensed under the MIT License. See LICENSE file for details.