1. Conduid
  2. Developer Tools
  3. Code Sandbox
MCP server · Developer Tools

Code Sandbox

MCP server: Code Sandbox

Unclaimed last commit a year ago devtools
42Fair

Scored 5 months ago · breakdown

About Code Sandbox

Code Sandbox is an MCP server published by chrishayuk in the Developer Tools category: mCP server: Code Sandbox.

The repository has 17 stars and 3 forks, with the last commit a year ago. Six months or more without a commit doesn't mean the server is broken, but check the open issues before depending on it in production.

Install

From source
git clone https://github.com/chrishayuk/mcp-code-sandbox

This server has no ConduID identity, so agent calls to it are not receipted. Pin the version you install and review the source before granting it credentials.

Ask AI

✦

Ask AI about Code Sandbox

Powered by Claude · Grounded in docs

I know everything about Code Sandbox. Ask me about installation, configuration, usage, or troubleshooting.

Security checks

Static checks derived from the repository yesterday. Heuristics on the source, not an audit.

  • ✓README presentThe repository ships documentation.
  • !License declaredNo license file. Legally you have no right to use it in a product.
  • ✓Tests presentA test suite exists in the repository.
  • ✓Dependencies pinnedA lockfile pins dependency versions.
  • ✓No dynamic code executionSource scan found no eval, dynamic Function, or shell execution.
  • !Scoped permissionsDoesn't declare a permission scope. Assume it can do anything its process can.

README

MCP Code Sandbox Server

An extensible Message Communication Protocol (MCP) server that provides secure code execution capabilities in isolated sandbox environments. This server follows the MCP standard, making it compatible with Claude for Desktop and other MCP clients.

Features

  • Create isolated sandbox environments for code execution
  • Execute Python code securely
  • Perform file operations (listing, reading, writing)
  • Install Python packages in the sandbox
  • Extensible architecture with abstracted code interpreter interface
  • Modular design with clean separation of concerns

Architecture

The server is built with a modular, extensible architecture:

Core Components

  • Abstract Interpreter Interface: Allows different code execution backends to be integrated
  • Sandbox Administration: Tools for creating and managing sandbox environments
  • Code Execution: Tools for running code and installing packages
  • File Operations: Tools for managing files within sandboxes

Project Structure

├── src/
│   └── sandbox/
│       ├── __pycache__/
│       ├── e2b/
│       │   ├── __pycache__/
│       │   ├── __init__.py
│       │   ├── e2b_file_interface.py
│       │   └── e2b_interpreter.py
│       ├── __init__.py
│       ├── code_interpreter.py
│       ├── file_interface.py
│       └── interpreter_factory.py
├── tools/
│   ├── __pycache__/
│   ├── __init__.py
│   ├── code_execution_tools.py
│   ├── file_tools.py
│   └── sandbox_tools.py
├── main.py
├── .env
├── .gitignore
├── .python-version
├── pyproject.toml
├── README.md
└── uv.lock

Prerequisites

  • Python 3.10 or higher
  • E2B API key (for the default E2B interpreter)

Installation

  1. Clone this repository:

    git clone https://github.com/yourusername/mcp-code-sandbox.git
    cd mcp-code-sandbox
    
  2. Set up a virtual environment:

    # Using venv
    python -m venv venv
    source venv/bin/activate  # On Windows: venv\Scripts\activate
    
    # Or using uv (recommended)
    uv init
    uv venv
    source .venv/bin/activate  # On Windows: .venv\Scripts\activate
    
  3. Install the required packages:

    # Using pip
    pip install fastmcp python-dotenv e2b-code-interpreter
    
    # Or using uv
    uv add fastmcp python-dotenv e2b-code-interpreter
    
  4. Configure environment variables:

    # Create a .env file with the following variables
    E2B_API_KEY=your_e2b_api_key_here
    INTERPRETER_TYPE=e2b  # Default, can be changed to other implemented interpreters
    

Usage

Running the Server Standalone

You can run the server directly from the command line:

python main.py

This will start the server using the stdio transport, making it compatible with Claude for Desktop.

Using with Claude for Desktop

  1. Make sure you have the latest version of Claude for Desktop installed

  2. Open your Claude for Desktop configuration file:

    • macOS: ~/Library/Application Support/Claude/claude_desktop_config.json
    • Windows: %APPDATA%\Claude\claude_desktop_config.json
  3. Add your code sandbox server configuration:

    {
      "mcpServers": {
        "code-sandbox": {
          "command": "python",
          "args": [
            "/ABSOLUTE/PATH/TO/main.py"
          ]
        }
      }
    }
    

    Or if you're using uv:

    {
      "mcpServers": {
        "code-sandbox": {
          "command": "uv",
          "args": [
            "--directory",
            "/ABSOLUTE/PATH/TO/PROJECT_DIRECTORY",
            "run",
            "main.py"
          ]
        }
      }
    }
    
  4. Save the file and restart Claude for Desktop

Available Tools

The server provides the following tools:

Sandbox Administration

  • create_sandbox: Create a new sandbox environment
  • close_sandbox: Close and clean up a sandbox
  • get_sandbox_status: Check status of sandboxes

Code Execution

  • execute_code: Run Python code in a sandbox
  • install_package: Install a Python package
  • create_run_close: All-in-one tool that creates a sandbox, runs code, and cleans up

File Operations

  • list_files: List files in the sandbox
  • read_file: Read the contents of a file
  • write_file: Write content to a file
  • upload_file: Upload a file to the sandbox

Extending with New Interpreters

The system is designed to be extensible. To add a new code interpreter:

  1. Create a new directory under src/sandbox/ for your interpreter implementation
  2. Implement the interfaces defined in src/sandbox/code_interpreter.py and src/sandbox/file_interface.py
  3. Add the new interpreter type to the src/sandbox/interpreter_factory.py
  4. Configure the environment variable INTERPRETER_TYPE to your new interpreter

Example of implementing a new interpreter:

# src/sandbox/my_backend/my_interpreter.py
from src.sandbox.code_interpreter import CodeInterpreter, ExecutionResult
from src.sandbox.file_interface import FileInterface

class MyFileInterface(FileInterface):
    # Implement the required methods
    
class MyInterpreter(CodeInterpreter):
    # Implement the required methods

# Update src/sandbox/interpreter_factory.py to include your new interpreter

Module Descriptions

Sandbox Core (src/sandbox/)

  • code_interpreter.py: Abstract base class for code interpreters
  • file_interface.py: Abstract interface for file operations
  • interpreter_factory.py: Factory for creating code interpreter instances

E2B Implementation (src/sandbox/e2b/)

  • e2b_interpreter.py: E2B implementation of the code interpreter
  • e2b_file_interface.py: E2B implementation of file operations

Tools (tools/)

  • sandbox_tools.py: Tools for sandbox administration
  • code_execution_tools.py: Tools for code execution
  • file_tools.py: Tools for file operations

Main Application

  • main.py: Main application entry point

Troubleshooting

If you encounter issues:

  • Make sure you have the correct API key for your chosen interpreter
  • Check the logs for detailed error messages
  • Verify that all required packages are installed
  • Ensure Claude for Desktop is configured with the correct path to your script

Security Considerations

  • The code execution happens in sandboxed environments for safety
  • Do not use this server to execute untrusted code in production environments
  • The server does not currently implement authentication - it should only be used in trusted environments

License

MIT License

README fetched from the repository yesterday. The original is authoritative.

Questions

About Code Sandbox

How do I install Code Sandbox?

No package has been published for this server, so install it from source: clone the repository and follow its README.

Is Code Sandbox safe to use with an AI agent?

Its trust score is 42 out of 100 (fair). It passes 4 of the 6 static security checks run so far; the failures are listed above. It has no ConduID identity yet, so agent calls to it are not receipted.

Is Code Sandbox still maintained?

The last commit was a year ago and 1 open issues. That's long enough that you should check whether the maintainer is responding to issues before depending on it.