Skip to main content
Glama
Node.js Sandbox MCP Server

Node.js Sandbox MCP Server

by ssdeanx

🐢🚀 Node.js Sandbox MCP Server

Node.js server implementing the Model Context Protocol (MCP) for running arbitrary JavaScript in ephemeral Docker containers with on‑the‑fly npm dependency installation.

Website Preview

👉 Look at the official website

Features

  • Start and manage isolated Node.js sandbox containers

  • Execute arbitrary shell commands inside containers

  • Install specified npm dependencies per job

  • Run ES module JavaScript snippets and capture stdout

  • Tear down containers cleanly

  • Detached Mode: Keep the container alive after script execution (e.g. for long-running servers)

Note: Containers run with controlled CPU/memory limits.

Related MCP server: Node.js Sandbox MCP Server

Explore Cool Use Cases

If you want ideas for cool and powerful ways to use this library, check out the use cases section on the website It contains a curated list of prompts, examples, and creative experiments you can try with the Node.js Sandbox MCP Server.

⚠️ Prerequisites

To use this MCP server, Docker must be installed and running on your machine.

Tip: Pre-pull any Docker images you'll need to avoid delays during first execution.

Example recommended images:

  • node:lts-slim

  • mcr.microsoft.com/playwright:v1.52.0-noble

  • alfonsograziano/node-chartjs-canvas:latest

Getting started

In order to get started with this MCP server, first of all you need to connect it to a client (for example Claude Desktop).

Once it's running, you can test that it's fully working with a couple of test prompts:

  • Validate that the tool can run:

    Create and run a JS script with a console.log("Hello World")

    This should run a console.log and in the tool response you should be able to see Hello World.

  • Validate that you can install dependencies and save files

    Create and run a JS script that generates a QR code for the URL `https://nodejs.org/en`, and save it as `qrcode.png` **Tip:** Use the `qrcode` package.

    This should create a file in your mounted directory (for example the Desktop) called "qrcode.png"

Usage with Claude Desktop

Add this to your claude_desktop_config.json: You can follow the Official Guide to install this MCP server

{
  "mcpServers": {
    "js-sandbox": {
      "command": "docker",
      "args": [
        "run",
        "-i",
        "--rm",
        "-v",
        "/var/run/docker.sock:/var/run/docker.sock",
        "-v",
        "$HOME/Desktop/sandbox-output:/root",
        "-e",
        "FILES_DIR=$HOME/Desktop/sandbox-output",
        "-e",
        "SANDBOX_MEMORY_LIMIT=512m", // optional
        "-e",
        "SANDBOX_CPU_LIMIT=0.75", // optional
        "alfonsograziano/node-code-sandbox-mcp"
      ]
    }
  }
}

or with NPX:

{
  "mcpServers": {
    "node-code-sandbox-mcp": {
      "type": "stdio",
      "command": "npx",
      "args": ["-y", "node-code-sandbox-mcp"],
      "env": {
        "FILES_DIR": "/Users/alfonsograziano/Desktop/node-sandbox",
        "SANDBOX_MEMORY_LIMIT": "512m", // optional
        "SANDBOX_CPU_LIMIT": "0.75" // optional
      }
    }
  }
}

Note: Ensure your working directory points to the built server, and Docker is installed/running.

Docker

Run the server in a container (mount Docker socket if needed), and pass through your desired host output directory as an env var:

# Build locally if necessary
# docker build -t alfonsograziano/node-code-sandbox-mcp .

docker run --rm -it \
  -v /var/run/docker.sock:/var/run/docker.sock \
  -v "$HOME/Desktop/sandbox-output":"/root" \
  -e FILES_DIR="$HOME/Desktop/sandbox-output" \
  -e SANDBOX_MEMORY_LIMIT="512m" \
  -e SANDBOX_CPU_LIMIT="0.5" \
  alfonsograziano/node-code-sandbox-mcp stdio

This bind-mounts your host folder into the container at the same absolute path and makes FILES_DIR available inside the MCP server.

Usage with VS Code

Quick install buttons (VS Code & Insiders):

Install js-sandbox-mcp (NPX) Install js-sandbox-mcp (Docker)

Manual configuration: Add to your VS Code settings.json or .vscode/mcp.json:

"mcp": {
    "servers": {
        "js-sandbox": {
            "command": "docker",
            "args": [
                "run",
                "-i",
                "--rm",
                "-v", "/var/run/docker.sock:/var/run/docker.sock",
                "-v", "$HOME/Desktop/sandbox-output:/root",
                "-e", "FILES_DIR=$HOME/Desktop/sandbox-output",
                "-e", "SANDBOX_MEMORY_LIMIT=512m",
                "-e", "SANDBOX_CPU_LIMIT=1",
                "alfonsograziano/node-code-sandbox-mcp"
              ]
        }
    }
}

API

Tools

run_js_ephemeral

Run a one-off JS script in a brand-new disposable container.

Inputs:

  • image (string, optional): Docker image to use (default: node:lts-slim).

  • code (string, required): JavaScript source to execute.

  • dependencies (array of { name, version }, optional): NPM packages and versions to install (default: []).

Behavior:

  1. Creates a fresh container.

  2. Writes your index.js and a minimal package.json.

  3. Installs the specified dependencies.

  4. Executes the script.

  5. Tears down (removes) the container.

  6. Returns the captured stdout.

  7. If your code saves any files in the current directory, these files will be returned automatically.

    • Images (e.g., PNG, JPEG) are returned as image content.

    • Other files (e.g., .txt, .json) are returned as resource content.

    • Note: the file saving feature is currently available only in the ephemeral tool.

Tip: To get files back, simply save them during your script execution.

Example Call:

{
  "name": "run_js_ephemeral",
  "arguments": {
    "image": "node:lts-slim",
    "code": "console.log('One-shot run!');",
    "dependencies": [{ "name": "lodash", "version": "^4.17.21" }],
  },
}

Example to save a file:

import fs from 'fs/promises';

await fs.writeFile('hello.txt', 'Hello world!');
console.log('Saved hello.txt');

This will return the console output and the hello.txt file.

sandbox_initialize

Start a fresh sandbox container.

  • Input:

    • image (string, optional, default: node:lts-slim): Docker image for the sandbox

    • port (number, optional): If set, maps this container port to the host

  • Output: Container ID string

sandbox_exec

Run shell commands inside the running sandbox.

  • Input:

    • container_id (string): ID from sandbox_initialize

    • commands (string[]): Array of shell commands to execute

  • Output: Combined stdout of each command

run_js

Install npm dependencies and execute JavaScript code.

  • Input:

    • container_id (string): ID from sandbox_initialize

    • code (string): JS source to run (ES modules supported)

    • dependencies (array of { name, version }, optional, default: []): npm package names → semver versions

    • listenOnPort (number, optional): If set, leaves the process running and exposes this port to the host (Detached Mode)

  • Behavior:

    1. Creates a temp workspace inside the container

    2. Writes index.js and a minimal package.json

    3. Runs npm install --omit=dev --ignore-scripts --no-audit --loglevel=error

    4. Executes node index.js and captures stdout, or leaves process running in background if listenOnPort is set

    5. Cleans up workspace unless running in detached mode

  • Output: Script stdout or background execution notice

sandbox_stop

Terminate and remove the sandbox container.

  • Input:

    • container_id (string): ID from sandbox_initialize

  • Output: Confirmation message

Usage Tips

  • Session-based tools (sandbox_initialize ➔ run_js ➔ sandbox_stop) are ideal when you want to:

    • Keep a long-lived sandbox container open.

    • Run multiple commands or scripts in the same environment.

    • Incrementally install and reuse dependencies.

  • One-shot execution with run_js_ephemeral is perfect for:

    • Quick experiments or simple scripts.

    • Cases where you don’t need to maintain state or cache dependencies.

    • Clean, atomic runs without worrying about manual teardown.

  • Detached mode is useful when you want to:

    • Spin up servers or long-lived services on-the-fly

    • Expose and test endpoints from running containers

Choose the workflow that best fits your use-case!

Build

Compile and bundle:

npm install
npm run build

License

MIT License

Permission is hereby granted, free of charge, to any person obtaining a copy of this software and associated documentation files (the "Software"), to deal in the Software without restriction, including without limitation the rights to use, copy, modify, merge, publish, distribute, sublicense, and/or sell copies of the Software, and to permit persons to whom the Software is furnished to do so, subject to the following conditions:

The above copyright notice and this permission notice shall be included in all copies or substantial portions of the Software.

THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.

Available Tools

7 tools
ai_generateC

Generate text using Google Gemini. Provide a prompt and optional model name.

Inputs: prompt (required), maxTokens, model.

get_dependency_typesB

Given an array of npm package names (and optional versions), fetch whether each package ships its own TypeScript definitions or has a corresponding @types/… package, and return the raw .d.ts text.

Useful whenwhen you're about to run a Node.js script against an unfamiliar dependency and want to inspect what APIs and types it exposes.

Inputs: dependencies (required).

run_jsA

Install npm dependencies and run JavaScript code inside a running sandbox container. After running, you must manually stop the sandbox to free resources. The code must be valid ESModules (import/export syntax). Best for complex workflows where you want to reuse the environment across multiple executions. When reading and writing from the Node.js processes, you always need to read from and write to the "./files" directory to ensure persistence on the mounted volume.

Inputs: code (required), container_id (required), dependencies, listenOnPort.

run_js_ephemeralA

Run a JavaScript snippet in a temporary disposable container with optional npm dependencies, then automatically clean up. The code must be valid ESModules (import/export syntax). Ideal for simple one-shot executions without maintaining a sandbox or managing cleanup manually. When reading and writing from the Node.js processes, you always need to read from and write to the "./files" directory to ensure persistence on the mounted volume. This includes images (e.g., PNG, JPEG) and other files (e.g., text, JSON, binaries).

Example:

import fs from "fs/promises";
await fs.writeFile("./files/hello.txt", "Hello world!");
console.log("Saved ./files/hello.txt");

Inputs: code (required), dependencies, image.

sandbox_execB

Execute one or more shell commands inside a running sandbox container. Requires a sandbox initialized beforehand.

Inputs: commands (required), container_id (required).

sandbox_initializeB

Start a new isolated Docker container running Node.js. Used to set up a sandbox session for multiple commands and scripts.

Inputs: image, port.

sandbox_stopA

Terminate and remove a running sandbox container. Should be called after finishing work in a sandbox initialized with sandbox_initialize.

Inputs: container_id (required).

Tool Schema Changelog

Recent tool additions, removals, and schema changes observed during successful MCP inspections.

Loading tool schema changelog…

    TDQS

    A3.6/5.0

    Scored across 7 tools

    Disambiguation4/5

    Most tools have distinct purposes: ai_generate is for text generation, get_dependency_types is for type inspection, and the sandbox tools (initialize, exec, stop) form a clear lifecycle. However, run_js and run_js_ephemeral overlap significantly in functionality—both run JavaScript code with dependencies—which could cause confusion despite differences in persistence and cleanup.

    Naming Consistency3/5

    Naming is mixed: ai_generate and get_dependency_types use snake_case, while sandbox_initialize, sandbox_exec, and sandbox_stop use a consistent 'sandbox_' prefix with snake_case. However, run_js and run_js_ephemeral deviate with a 'run_' prefix and lack the 'sandbox_' pattern, creating inconsistency. The verb styles vary (e.g., 'get', 'run', 'sandbox_exec'), reducing predictability.

    Tool Count5/5

    With 7 tools, the count is well-scoped for a Node.js sandbox server. It covers core areas: AI text generation, dependency type inspection, JavaScript execution (both persistent and ephemeral), and sandbox management (initialize, execute commands, stop). Each tool earns its place without feeling bloated or insufficient.

    Completeness4/5

    The toolset provides strong coverage for Node.js sandbox operations, including sandbox lifecycle management and code execution. A minor gap exists: there's no tool for inspecting or managing files within the sandbox beyond the implied './files' directory usage, which agents might need to work around. Otherwise, core workflows are well-supported.

    Related MCP Connectors

    Related MCP Servers

    • A
      license
      Not graded
      quality
      C
      maintenance
      Facilitates isolated code execution within Docker containers, enabling secure multi-language script execution and integration with language models like Claude via the Model Context Protocol.
      5
      MIT
    • F
      license
      Not graded
      quality
      D
      maintenance
      Enables running arbitrary JavaScript code in isolated Docker containers with on-the-fly npm dependency installation, supporting both ephemeral one-shot executions and persistent sandbox environments.
      156
      -
    • F
      license
      A
      quality
      D
      maintenance
      A secure Node.js execution environment that allows coding agents and LLMs to run JavaScript dynamically, install NPM packages, and retrieve results while adhering to the Model Control Protocol.
      7
      4
      -
    • A
      license
      Not graded
      quality
      D
      maintenance
      Enables LLMs to safely execute code in isolated Docker containers with resource limits and security controls, supporting session management and automatic dependency installation.
      MIT