Node.js Sandbox MCP Server
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.
README
🐢🚀 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.
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
Note: Containers run with controlled CPU/memory limits.
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:20-slim).code(string, required): JavaScript source to execute.dependencies(array of{ name, version }, optional): NPM packages and versions to install (default:[]).
Behavior:
- Creates a fresh container.
- Writes your
index.jsand a minimalpackage.json. - Installs the specified dependencies.
- Executes the script.
- Tears down (removes) the container.
- Returns the captured stdout.
Example Call:
{
"name": "run_js_ephemeral",
"arguments": {
"image": "node:20-slim",
"code": "console.log('One-shot run!');",
"dependencies": [{ "name": "lodash", "version": "^4.17.21" }]
}
}
sandbox_initialize
Start a fresh sandbox container.
- Input:
image(string, optional, default:node:20-slim): Docker image for the sandbox
- Output: Container ID string
sandbox_exec
Run shell commands inside the running sandbox.
- Input:
container_id(string): ID fromsandbox_initializecommands(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 fromsandbox_initializecode(string): JS source to run (ES modules supported)dependencies(array of{ name, version }, optional, default:[]): npm package names → semver versions
- Behavior:
- Creates a temp workspace inside the container
- Writes
index.jsand a minimalpackage.json - Runs
npm install --omit=dev --ignore-scripts --no-audit --loglevel=error - Executes
node index.jsand captures stdout - Cleans up workspace
- Output: Script stdout
sandbox_stop
Terminate and remove the sandbox container.
- Input:
container_id(string): ID fromsandbox_initialize
- Output: Confirmation message
Usage with Claude Desktop
Add this to your claude_desktop_config.json:
{
"mcpServers": {
"js-sandbox": {
"command": "node",
"args": ["dist/server.js", "stdio"],
"cwd": "/path/to/js-sandbox-mcp"
}
}
}
Note: Ensure your working directory points to the built server, and Docker is installed/running.
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_ephemeralis 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.
Choose the workflow that best fits your use-case!
Docker
Run the server in a container (mount Docker socket if needed):
# Build and publish your image locally if needed
# docker build -t alfonsograziano/node-code-sandbox-mcp .
# Run with stdio transport
docker run --rm -it \
-v /var/run/docker.sock:/var/run/docker.sock \
alfonsograziano/node-code-sandbox-mcp stdio
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",
"alfonsograziano/node-code-sandbox-mcp"
],
}
}
}
Usage with Claude Desktop
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",
"alfonsograziano/node-code-sandbox-mcp"
]
}
}
}
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.
推荐服务器
Baidu Map
百度地图核心API现已全面兼容MCP协议,是国内首家兼容MCP协议的地图服务商。
Playwright MCP Server
一个模型上下文协议服务器,它使大型语言模型能够通过结构化的可访问性快照与网页进行交互,而无需视觉模型或屏幕截图。
Magic Component Platform (MCP)
一个由人工智能驱动的工具,可以从自然语言描述生成现代化的用户界面组件,并与流行的集成开发环境(IDE)集成,从而简化用户界面开发流程。
Audiense Insights MCP Server
通过模型上下文协议启用与 Audiense Insights 账户的交互,从而促进营销洞察和受众数据的提取和分析,包括人口统计信息、行为和影响者互动。
VeyraX
一个单一的 MCP 工具,连接你所有喜爱的工具:Gmail、日历以及其他 40 多个工具。
graphlit-mcp-server
模型上下文协议 (MCP) 服务器实现了 MCP 客户端与 Graphlit 服务之间的集成。 除了网络爬取之外,还可以将任何内容(从 Slack 到 Gmail 再到播客订阅源)导入到 Graphlit 项目中,然后从 MCP 客户端检索相关内容。
Kagi MCP Server
一个 MCP 服务器,集成了 Kagi 搜索功能和 Claude AI,使 Claude 能够在回答需要最新信息的问题时执行实时网络搜索。
e2b-mcp-server
使用 MCP 通过 e2b 运行代码。
Neon MCP Server
用于与 Neon 管理 API 和数据库交互的 MCP 服务器
Exa MCP Server
模型上下文协议(MCP)服务器允许像 Claude 这样的 AI 助手使用 Exa AI 搜索 API 进行网络搜索。这种设置允许 AI 模型以安全和受控的方式获取实时的网络信息。