OpenSandbox
Secure, Fast, and Extensible Sandbox runtime for AI agents.
OpenSandbox MCP server provides secure sandbox execution capabilities for AI agents through the Model Context Protocol. It enables MCP clients such as Claude Code and Cursor to create isolated sandbox environments, execute commands, and perform file operations within Docker or Kubernetes runtimes. The MCP server exposes OpenSandbox's sandbox lifecycle APIs—including creation, command execution, and text file manipulation—allowing AI coding assistants to safely run arbitrary code and scripts. Developers can configure it to connect to local or remote OpenSandbox instances for both testing and production agent workflows.
Key Features
Use Cases
- 01Running AI coding agents (Claude Code, Gemini CLI, Qwen Code) with isolated code execution
- 02Executing untrusted code safely from LLM-generated scripts or user prompts
- 03Browser automation and web scraping with sandboxed Chrome or Playwright instances
- 04Agent evaluation and reinforcement learning training with one sandbox per trial
- 05LangGraph workflows that spawn ephemeral sandboxes for task execution with retry logic
- 06Remote development environments using code-server (VS Code Web) inside sandboxes
Related MCP Servers
View moremarkitdown
Python tool for converting files and office documents to Markdown.
firecrawl
The context API to search, scrape, and interact with the web at scale. 🔥
prompts.chat
f.k.a. Awesome ChatGPT Prompts. Share, discover, and collect prompts from the community. Free and open source — self-host for your organization with complete privacy.
langflow
Langflow is a powerful tool for building and deploying AI-powered agents and workflows.
OpenSandbox — FAQ
What is the OpenSandbox MCP server?+
The OpenSandbox MCP server is a Model Context Protocol implementation that lets AI clients like Claude Code and Cursor create secure sandbox environments, run commands, and manipulate files through standardized MCP tools.
How do I install the OpenSandbox MCP server?+
Install via pip with 'pip install opensandbox-mcp', then add a JSON configuration entry to your MCP client's settings specifying the command 'opensandbox-mcp' with arguments pointing to your OpenSandbox server domain and protocol. You must also have an OpenSandbox server instance running (locally via Docker or remotely on Kubernetes).
Which AI clients work with the OpenSandbox MCP server?+
The MCP server works with any MCP-capable client including Claude Code, Cursor, and other tools that support the Model Context Protocol stdio interface.
Do I need API keys or a running server to use the OpenSandbox MCP server?+
You need a running OpenSandbox server instance (local or remote) and must configure the MCP server with the server's domain, protocol (http/https), and optionally an API key if authentication is enabled on your OpenSandbox server.
Is the OpenSandbox MCP server free and open source?+
Yes, OpenSandbox is fully open source under the Apache 2.0 License. All components including the MCP server, SDKs, CLI, and runtimes are freely available on GitHub.
What are the prerequisites for running OpenSandbox locally?+
Docker is required for local sandbox execution, and Python 3.10 or later is needed to run the OpenSandbox server and MCP server components. For production Kubernetes deployments, a configured Kubernetes cluster is required.
How do I install OpenSandbox?+
Open the source repository on GitHub and follow its README. OpenSandbox is a mcp server — MCP Agents Market links you directly to the official repo.
Is OpenSandbox free?+
OpenSandbox is an open-source project hosted on GitHub. Check the repository for its license and any usage requirements.